@leadbay/mcp 0.38.0 → 0.38.2

package/dist/http-server.js

@@ -38,7 +38,7 @@ What Leadbay does need is anything that changes **who it should find**. When the
 
 Build me a Leadbay campaign from scratch{{arg:campaign_name_paren}} \u2014 a cohort of **{{arg:count_or_default}}** fully-actionable leads: each in-ICP, high \`ai_agent_lead_score\`, AND with a reachable buyer contact. {{arg:audience_block}} {{arg:job_titles_block}}
 
-**Run this end-to-end, autonomously, without pausing.** Do NOT stop to confirm the audience, do NOT stop to confirm the enrichment spend, do NOT ask me to pick, and do NOT stop to hand off \u2014 just keep discovering, qualifying, enriching, and swapping until the cohort holds **{{arg:count_or_default}}** leads that each meet EVERY requirement (in-ICP, high \`ai_agent_lead_score\`, and a reachable target-title contact whose email/phone actually landed). The ONLY reasons to stop short: the lens genuinely can't supply that many buyer-ready in-ICP leads, or enrichment quota is exhausted (a backend 429). In those cases, finish with whatever you locked and tell me plainly how many you got and why it stopped. Enrichment consumes quota, not credits \u2014 never pre-refuse on a credit balance.
+**Run this end-to-end, autonomously, without pausing.** Do NOT stop to confirm the audience, do NOT stop to confirm the enrichment, do NOT ask me to pick, and do NOT stop to hand off \u2014 just keep discovering, qualifying, enriching, and swapping until the cohort holds **{{arg:count_or_default}}** leads that each meet EVERY requirement (in-ICP, high \`ai_agent_lead_score\`, and a reachable target-title contact whose email/phone actually landed). The ONLY reasons to stop short: the lens genuinely can't supply that many buyer-ready in-ICP leads, or enrichment quota is exhausted (a backend 429). In those cases, finish with whatever you locked and tell me plainly how many you got and why it stopped. Enrichment consumes quota, not credits \u2014 never pre-refuse on a credit balance.
 
 GATE \u2014 DEFER TO TOOL RENDERING. When you call a Leadbay composite that ships its own RENDERING block (every composite in 0.9.0+ does), render the response using that block's recipe verbatim \u2014 score bars, glyph palette, column order, hide-list, link priorities, all of it. Do NOT substitute prose, a numbered list, or a different column structure even when an orchestrating prompt's body suggests alternate framing. Prompt-specific commentary (motivational nudges, summaries, next-action recommendations) belongs ABOVE or BELOW the canonical table, never in place of it.
 
@@ -76,7 +76,7 @@ If \`pull_leads\` itself fails and you have no prior batch, then yes \u2014 retr
 Leadbay has no cancel. Once \`leadbay_enrich_titles\`, \`leadbay_bulk_qualify_leads\`,
 \`leadbay_import_leads\` or \`leadbay_import_and_qualify\` has returned a launched or
 running result, that work is queued on Leadbay and runs to completion, and the
-quota it costs is already committed. A discovery, preview or \`dry_run\` result
+quota it uses is already committed. A discovery, preview or \`dry_run\` result
 launched nothing and is not covered here.
 
 The user cancelling in the chat, a request timeout, or a closed stream stops YOUR
@@ -84,7 +84,7 @@ waiting, never the job. \`cancelled: true\` means we stopped watching, not that
 work stopped. What to do next depends on what you are holding:
 
 - **A handle.** Poll the status tool with it, and do not launch the work that
-  handle covers a second time \u2014 that spends the quota again on the same rows.
+  handle covers a second time \u2014 that uses the quota again on the same rows.
   \`leadbay_import_status\` takes \`importIds\`, so pass the values of \`import_ids\`
   under that name. A qualification started by \`leadbay_import_and_qualify\` has no
   notification of its own: resume it with
@@ -98,7 +98,7 @@ work stopped. What to do next depends on what you are holding:
   finished. Calling the same tool again with the same arguments will usually hand
   back the job already launched rather than starting a second one, but that guard
   is in-memory, five minutes, and per process, so it is best-effort \u2014 say what you
-  are about to re-run before you spend the user's quota on it.
+  are about to re-run before you use the user's quota on it.
 
 
 
@@ -197,7 +197,7 @@ If I named specific leads, seed with those (still apply the Phase 3 buyer-covera
 
 # PHASE 3 \u2014 ENRICH THE RIGHT CONTACTS (load-bearing)
 
-This is the phase that decides whether the campaign is worth a salesperson's time. Contacts aren't attached by default and enrichment is paid \u2014 so spend it ONLY on the people who would actually **buy what I sell**, at the target titles, not on whoever is most senior.
+This is the phase that decides whether the campaign is worth a salesperson's time. Contacts aren't attached by default and enrichment uses quota \u2014 so use it ONLY on the people who would actually **buy what I sell**, at the target titles, not on whoever is most senior.
 
 **Step A \u2014 settle the target titles / buyer persona.**
 
@@ -223,9 +223,9 @@ Call \`leadbay_recall_ordered_titles({leadIds, lensId})\` and \`leadbay_enrich_t
 
 Tell me what you swapped in one line ("dropped Corbett + RBS \u2014 ops-only; swapped in Acme + Globex which have Sales VPs").
 
-**Step C \u2014 enrich (NO confirm gate \u2014 just spend).** You do NOT need my permission: I authorized this spend by asking for the campaign. Do NOT call \`ask_user_input_v0\`, do NOT ask "enrich these N now?", do NOT wait. State the persona + titles + "enriching {enrichable_contacts} contacts (email + phone, consumes quota)" in one line for the record, then immediately launch: \`leadbay_enrich_titles({leadIds, lensId, titles:[...chosen], email:true, phone:true})\`. Enrich up to {{arg:count_or_default}} best target-title contacts. Do NOT quote a "credits" figure or refuse on a credit balance \u2014 the only real limit is quota (a backend 429). If a 429 stops you mid-run, keep the leads already enriched, note how many landed, and continue to Phase 4 with those.
+**Step C \u2014 enrich (NO confirm gate \u2014 just launch).** You do NOT need my permission: I authorized this enrichment by asking for the campaign. Do NOT call \`ask_user_input_v0\`, do NOT ask "enrich these N now?", do NOT wait. State the persona + titles + "enriching {enrichable_contacts} contacts (email + phone, consumes quota)" in one line for the record, then immediately launch: \`leadbay_enrich_titles({leadIds, lensId, titles:[...chosen], email:true, phone:true})\`. Enrich up to {{arg:count_or_default}} best target-title contacts. Do NOT quote a "credits" figure or refuse on a credit balance \u2014 the only real limit is quota (a backend 429). If a 429 stops you mid-run, keep the leads already enriched, note how many landed, and continue to Phase 4 with those.
 
-**Step D \u2014 poll + count only landed.** Poll \`leadbay_bulk_enrich_status\` until done (enrichment can take several minutes \u2014 keep polling, don't render an empty sheet prematurely). Once \`all_done\`, call \`leadbay_account_status\` and show my refreshed quota so I see what the run consumed. A lead only counts toward the {{arg:count_or_default}} once its target-title contact actually landed (email/phone present); if some came back empty, swap + enrich replacements (loop back to Step B.5) until the cohort is genuinely {{arg:count_or_default}} deep or the lens is exhausted.
+**Step D \u2014 poll + count only landed.** Poll \`leadbay_bulk_enrich_status\` until done (enrichment can take several minutes \u2014 keep polling, don't render an empty sheet prematurely). A lead only counts toward the {{arg:count_or_default}} once its target-title contact actually landed (email/phone present); if some came back empty, swap + enrich replacements (loop back to Step B.5) until the cohort is genuinely {{arg:count_or_default}} deep or the lens is exhausted.
 
 # PHASE 4 \u2014 CREATE THE CAMPAIGN
 
@@ -248,7 +248,7 @@ Building a campaign is NOT outreaching \u2014 do not send anything and do not ca
 # Iron laws
 
 - **Run to the goal, autonomously.** Keep discovering \u2192 qualifying \u2192 enriching \u2192 swapping until the cohort holds {{arg:count_or_default}} leads that are ALL in-ICP, high-score, and buyer-covered \u2014 or the lens is genuinely exhausted. Do NOT stop early, do NOT ask me to pick, do NOT hand off mid-flow.
-- **No confirm gates. No pauses.** Do NOT confirm the audience switch, and do NOT confirm the enrichment spend (no \`ask_user_input_v0\` before enriching) \u2014 asking for the campaign IS the authorization. The only acceptable stops are lens exhaustion or a backend 429.
+- **No confirm gates. No pauses.** Do NOT confirm the audience switch, and do NOT confirm the enrichment (no \`ask_user_input_v0\` before enriching) \u2014 asking for the campaign IS the authorization. The only acceptable stops are lens exhaustion or a backend 429.
 - Enrichment targets MY buyer titles \u2014 the people who would actually buy what *I* sell (my given titles, or the persona derived from my product/ICP) \u2014 NOT generic seniority. For a sales/prospecting tool that means the revenue org; a Director of Operations, COO, or logistics manager is useless no matter how senior.
 - Selection is DATA-DRIVEN (\`leadbay_recall_ordered_titles\` + \`leadbay_enrich_titles\` discovery) but FILTERED to the target titles \u2014 never blindly repeat past-enriched or suggested titles that don't match who buys my product.
 - The FINAL cohort must be all buyer-ready: a lead counts only once its target-title contact actually landed. Drop/swap + re-enrich any lead with no reachable buyer rather than shipping it empty.
@@ -298,7 +298,7 @@ If \`pull_leads\` itself fails and you have no prior batch, then yes \u2014 retr
 Leadbay has no cancel. Once \`leadbay_enrich_titles\`, \`leadbay_bulk_qualify_leads\`,
 \`leadbay_import_leads\` or \`leadbay_import_and_qualify\` has returned a launched or
 running result, that work is queued on Leadbay and runs to completion, and the
-quota it costs is already committed. A discovery, preview or \`dry_run\` result
+quota it uses is already committed. A discovery, preview or \`dry_run\` result
 launched nothing and is not covered here.
 
 The user cancelling in the chat, a request timeout, or a closed stream stops YOUR
@@ -306,7 +306,7 @@ waiting, never the job. \`cancelled: true\` means we stopped watching, not that
 work stopped. What to do next depends on what you are holding:
 
 - **A handle.** Poll the status tool with it, and do not launch the work that
-  handle covers a second time \u2014 that spends the quota again on the same rows.
+  handle covers a second time \u2014 that uses the quota again on the same rows.
   \`leadbay_import_status\` takes \`importIds\`, so pass the values of \`import_ids\`
   under that name. A qualification started by \`leadbay_import_and_qualify\` has no
   notification of its own: resume it with
@@ -320,7 +320,7 @@ work stopped. What to do next depends on what you are holding:
   finished. Calling the same tool again with the same arguments will usually hand
   back the job already launched rather than starting a second one, but that guard
   is in-memory, five minutes, and per process, so it is best-effort \u2014 say what you
-  are about to re-run before you spend the user's quota on it.
+  are about to re-run before you use the user's quota on it.
 
 
 
@@ -329,7 +329,7 @@ work stopped. What to do next depends on what you are holding:
 If you're resuming an interrupted session (you see a previous Phase already completed in your task list, or the user says "continue" / "continue from where you left off"), do NOT restart from Phase 1. Re-read the active \`lensId\` and your last completed phase from prior context, then resume from the next phase. If you genuinely have no state, restart from Phase 1.
 
 # PHASE 1 \u2014 STATE
-Call \`leadbay_account_status\` to see what quota I have left and which lens is active. Note the remaining \`ai_rescore_remaining\` and \`web_fetch_remaining\` budgets \u2014 Phase 4 enrichment depends on them.
+Call \`leadbay_account_status\` to see what quota I have left and which lens is active. Note the remaining \`ai_rescore_remaining\` and \`web_fetch_remaining\` budgets \u2014 Phase 4 enrichment depends on them. This is for your planning: don't show me the quota unless a window is exhausted.
 
 # PHASE 2 \u2014 FRESH BATCH
 Call \`leadbay_pull_leads\` to get today's fresh batch. Capture \`response.lens.id\` (the response nests it under \`lens\`). **Use it as an explicit \`lensId\` argument on every subsequent Leadbay call this session** \u2014 including any re-pulls, bulk qualifies, or research calls that accept it. (See Rule 1 above \u2014 a mid-session lens shift discards your top-10 work.)
@@ -560,7 +560,7 @@ If \`pull_leads\` itself fails and you have no prior batch, then yes \u2014 retr
 Leadbay has no cancel. Once \`leadbay_enrich_titles\`, \`leadbay_bulk_qualify_leads\`,
 \`leadbay_import_leads\` or \`leadbay_import_and_qualify\` has returned a launched or
 running result, that work is queued on Leadbay and runs to completion, and the
-quota it costs is already committed. A discovery, preview or \`dry_run\` result
+quota it uses is already committed. A discovery, preview or \`dry_run\` result
 launched nothing and is not covered here.
 
 The user cancelling in the chat, a request timeout, or a closed stream stops YOUR
@@ -568,7 +568,7 @@ waiting, never the job. \`cancelled: true\` means we stopped watching, not that
 work stopped. What to do next depends on what you are holding:
 
 - **A handle.** Poll the status tool with it, and do not launch the work that
-  handle covers a second time \u2014 that spends the quota again on the same rows.
+  handle covers a second time \u2014 that uses the quota again on the same rows.
   \`leadbay_import_status\` takes \`importIds\`, so pass the values of \`import_ids\`
   under that name. A qualification started by \`leadbay_import_and_qualify\` has no
   notification of its own: resume it with
@@ -582,7 +582,7 @@ work stopped. What to do next depends on what you are holding:
   finished. Calling the same tool again with the same arguments will usually hand
   back the job already launched rather than starting a second one, but that guard
   is in-memory, five minutes, and per process, so it is best-effort \u2014 say what you
-  are about to re-run before you spend the user's quota on it.
+  are about to re-run before you use the user's quota on it.
 
 
 
@@ -706,10 +706,10 @@ Mirror the Leadbay web quota widget: three windows side by side \u2014 **Daily**
 $ cap** figure, with a per-resource usage breakdown underneath. **Never speak in
 raw "credits"** for quota \u2014 the unit is a percentage and a dollar spend.
 
-**Include the quota whenever it is readable** \u2014 as part of the default account
-answer, even when the user only asked "what account am I connected to?". The
-sole reason to omit it is the silence gate below (unreadable quota, or an
-unlimited account); it is NOT gated on the user explicitly asking for quota.
+**Show the quota only when it matters** \u2014 when the user asks about their quota,
+usage or account status, or when a window is exhausted and blocks what they
+asked for. A plain "what account am I connected to?" is answered with user +
+org alone. Even then, the silence gate below comes first.
 
 **Silence gate (check FIRST).** Render NOTHING about quota when any of these
 holds \u2014 do not mention quota at all, do not say "unreadable", never tell the user
@@ -769,7 +769,7 @@ days"), computed against now \u2014 mirroring the widget's "r\xE9initialis\xE9 d
 raw value is an ISO-8601 timestamp.
 
 **Top-up (optional, subordinate).** When \`quota.topup\` is present, you MAY add one
-small line below the windows: \`Top-up: $<remaining_cents/100> of $<total_credit_cents/100> left\`.
+small line below the windows: \`Top-up: $<(remaining_cents / 100).toFixed(2)> of $<(total_credit_cents / 100).toFixed(2)> left\`.
 Keep it secondary \u2014 the three window gauges are the headline. Omit when null.
 
 **Legend** (once, below): \`\` \`\u25B0\` used \xB7 \`\u25B1\` remaining \`\`.
@@ -938,9 +938,9 @@ Say plainly that this only **drafts** \u2014 nothing is sent, and they see it fi
 On click: call \`leadbay_prepare_outreach\` with \`leadId\` = the top lead's id,
 **and nothing else**.
 
-**This gate spends NOTHING. Never pass \`enrich: true\`** \u2014 that launches a paid
+**This gate uses NO quota. Never pass \`enrich: true\`** \u2014 that launches a
 contact reveal off the back of a *draft* click. They agreed to see an email
-written, not to spend. GATE 4 is where the reveal gets asked for, on its own
+written, not to reveal anyone. GATE 4 is where the reveal gets asked for, on its own
 terms.
 
 \`recommended_contact\` comes back in its post-enrichment shape with \`email\` and
@@ -974,7 +974,7 @@ pitching whoever answers the switchboard \u2014 the difference between a
 conversation and a dead end.
 
 Say plainly that the first look is **free**, and that revealing the contact
-costs credits and needs their say-so.
+uses a little of their plan's quota and needs their say-so.
 
 **First, check \`leadbay_enrich_titles\` is in your tool set.** On a read-only
 deployment it is not registered, and a gate whose tool cannot run is a dead
@@ -987,23 +987,23 @@ does nothing.
 
 This gate runs in **TWO BEATS**. Do not collapse them.
 
-## BEAT 1 \u2014 the free look (spends nothing)
+## BEAT 1 \u2014 the free look (uses no quota)
 
 On click: call \`leadbay_enrich_titles\` with \`leadIds\` = **the one lead you
 drafted for at GATE 3** and \`lensId\` = the pinned lens id.
 
-**This call must spend NOTHING.** Omit \`titles\` entirely: that returns
+**This call must use NO quota.** Omit \`titles\` entirely: that returns
 \`mode:"discover"\`, the free preview of which job titles exist at that company.
 Do NOT pass \`titles\`, \`confirm=true\`, \`email=true\` or \`phone=true\` on this call
-\u2014 any one of them launches the paid reveal before the user has chosen anything.
+\u2014 any one of them launches the reveal before the user has chosen anything.
 
-Present the discovered titles and say plainly: "nothing spent yet."
+Present the discovered titles and say plainly: "nothing revealed yet."
 
-## BEAT 2 \u2014 reveal the person the draft is for (spends credits)
+## BEAT 2 \u2014 reveal the person the draft is for (uses quota)
 
-Name the title the GATE 3 draft is addressed to, and tell them the cost
-**before** they decide: one credit per contact revealed \u2014 here that's **one
-contact, one credit**. Then ask them to confirm.
+Name the title the GATE 3 draft is addressed to, and tell them **before** they
+decide that revealing **one contact** uses a little of their plan's quota. No
+amount, no price. Then ask them to confirm.
 
 **Wait for an explicit confirmation.** Silence is not consent, and neither is
 "they clicked the gate earlier" \u2014 the gate click bought the free look, not the
@@ -1011,13 +1011,13 @@ reveal.
 
 Once confirmed, call \`leadbay_enrich_titles\` AGAIN with
 \`leadIds: [<the drafted lead's id>]\` \u2014 **the array, always, even for one lead**
-\u2014 plus the chosen \`titles\`, \`confirm: true\` and \`email: true\`. That's the real,
-paid reveal.
+\u2014 plus the chosen \`titles\`, \`confirm: true\` and \`email: true\`. That's the real
+reveal.
 
 \`leadIds\` is the only key this tool reads for scope. A singular \`leadId\` is not
 a parameter: it is silently ignored, and the call then falls back to the
 account's **default wishlist selection** while \`confirm\`/\`email\` are set \u2014 so
-it would reveal and charge for the whole batch instead of the one lead the user
+it would reveal the whole batch instead of the one lead the user
 agreed to.
 
 It returns a \`notification_id\` and runs async \u2014 poll \`leadbay_bulk_enrich_status\`
@@ -1026,8 +1026,8 @@ count plateaus across a few spaced polls. Then report the contact that actually
 resolved: name, title, and the email/phone that came back. Contacts sometimes
 don't resolve; say so honestly rather than implying success.
 
-**Then close the loop** \u2014 one line: one credit per contact revealed, so this
-cost one. And say the thing that makes it land: the draft from GATE 3 now has a
+**Then close the loop** \u2014 one line: the reveal used a little of their plan's
+quota, never a price. And say the thing that makes it land: the draft from GATE 3 now has a
 real person and a real address to go to. This is the moment GATE 1's quota
 numbers stop being abstract, because they just watched them move and got
 something for it. Don't turn it into a pricing pitch.
@@ -1387,12 +1387,12 @@ Call \`leadbay_find_new_leads\` with the seed, \`filters\`, \`count\`,
 from the ask + the ARCHETYPE + today's date. \`count\` is the TOTAL I asked
 for, not a per-search number: with two archetypes and a request for 10,
 split it (5 + 5, or whatever weighting fits my ask) rather than sending 10
-to each \u2014 otherwise I get 20 leads and, on the paid pass, pay for 20.
+to each \u2014 otherwise I get 20 leads and, on the qualifying pass, use quota on 20.
 
 When you RETRY a search \u2014 it timed out, or the job is still live \u2014 reuse the
 \`request_id\` you already sent, verbatim. Do not recompute it: rederiving from
 "today's date" after midnight yields a new key, the backend cannot dedupe, and
-a second paid, novelty-claiming search launches. Roll the date only when I am
+a second quota-using, novelty-claiming search launches. Roll the date only when I am
 genuinely asking for a new batch. The archetype component is not
 optional: \`request_id\` is the idempotency key, so two archetype searches
 sharing one id dedupe to the same job and the second archetype is never
@@ -1407,52 +1407,54 @@ kind of companies I asked for?
 - **Off-profile or empty** (terminal) \u2192 read \`funnel\` +
   \`explain.scope_notes\`, tell me what went wrong in one line (wrong
   archetype? too narrow a filter? thin universe?), reshape the seed or
-  filters, and retry under a NEW request_id. Reshaping is free; do not pay
-  to explore a bad seed.
+  filters, and retry under a NEW request_id. Reshaping is free; do not use
+  quota to explore a bad seed.
 
-# PHASE 4 \u2014 PAID DEPTH (only with my explicit go-ahead)
+# PHASE 4 \u2014 QUALIFIED DEPTH (uses my plan's quota; only with my explicit go-ahead)
 
 When I want qualification evidence and/or reachable contacts:
-1. Quote first: \`dry_run: true\` on the tool you will actually run, with the
-   exact flags I asked for, and tell me the worst-case cost in plain money.
+1. Check first: \`dry_run: true\` on the tool you will actually run, with the
+   exact flags I asked for. Tell me what will run (how many companies, which
+   channels) and that it uses my plan's quota. No amounts and no money: my plan
+   or top-up covers this, and a price reads as a bill.
    The two tools take DIFFERENT flags \u2014 passing the wrong one is rejected
    outright (\`additionalProperties: false\`):
    - \`leadbay_qualify_leads\`: \`qualify: true\`, \`contact_titles\`,
      \`title_gate\`, \`channels\`, \`max_cost\`. **No \`min_ai_score\`.**
    - \`leadbay_find_new_leads\`: the same, PLUS \`min_ai_score\` and \`count\`.
 2. On my go-ahead, prefer feeding the free preview's deliveries to
-   \`leadbay_qualify_leads\` (\`prior_deliveries: {job_id}\`) \u2014 one paid pass PER
+   \`leadbay_qualify_leads\` (\`prior_deliveries: {job_id}\`) \u2014 one qualifying pass PER
    preview job when Phase 3 ran several archetypes, or merge their delivered
    refs into a single \`lead_refs\` call. Never qualify just the first job and
    call it done: the other archetypes are part of what I asked for. It only
-   spends on
-   companies already known to match. Paid calls need \`confirm: true\`; without
+   uses quota on
+   companies already known to match. Calls that use quota need \`confirm: true\`; without
    it the tool withholds the submit and hands back a quote instead of
-   spending. That applies to \`leadbay_find_new_leads\` too whenever you set
+   running. That applies to \`leadbay_find_new_leads\` too whenever you set
    \`qualify: true\` or ask for channels.
 
    If the preview delivered FEWER than I asked for, do both halves and do not
    conflate them: qualify what the preview already found, and run the fresh
    search only for the SHORTFALL \u2014 \`count\` = what is still missing, never the
    original number, under a NEW \`request_id\`. Reusing the preview's id dedupes
-   the paid submit back into the free job; keeping the original count buys a
+   the qualifying submit back into the free job; keeping the original count runs a
    whole second batch, because \`novelty: org\` already excludes everything the
    preview delivered.
 
-   The same arithmetic applies AFTER the paid pass. A full-count preview can
+   The same arithmetic applies AFTER the qualifying pass. A full-count preview can
    still end short once qualification disqualifies rows or a strict title /
    channel match misses: what I asked for is n QUALIFIED, CONTACTABLE leads,
    not n examined. Count the delivered-and-callable rows; if they fall short,
-   tell me the gap in one line and offer to top it up \u2014 another shortfall-sized
-   search under a NEW \`request_id\`, quoted first like any paid run. Never
-   silently hand back fewer than I asked for and paid toward.
+   tell me the gap in one line and offer to fill the gap \u2014 another shortfall-sized
+   search under a NEW \`request_id\`, confirmed first like any run that uses
+   quota. Never silently hand back fewer than I asked for.
 
-   Pass the leads already EXAMINED-AND-REJECTED into that top-up's
+   Pass the leads already EXAMINED-AND-REJECTED into that search's
    \`exclude_lead_ids\` \u2014 disqualified and skipped, from both the preview and
-   the paid pass. \`novelty: org\` already excludes prior DELIVERIES, so
+   the qualifying pass. \`novelty: org\` already excludes prior DELIVERIES, so
    delivered ids are redundant there; the rejected ones are exactly what it
-   misses, and without them the top-up re-picks the same misses and charges
-   again to close no gap. **\`exclude_lead_ids\` caps at 500** \u2014 a wide
+   misses, and without them that search re-picks the same misses and uses
+   quota again to close no gap. **\`exclude_lead_ids\` caps at 500** \u2014 a wide
    \`exploration_cap\` can examine more than that, so send the most recent 500
    rejects rather than an over-long list the tool refuses outright.
 3. While the job runs, poll with \`leadbay_lead_job_status\`
@@ -1468,7 +1470,7 @@ best fit is under 30, say "weak matches only" and propose reshaping before
 showing more than 3.
 
 Render per the lead-delivery table, then ALWAYS the funnel line: matched /
-examined / qualified / disqualified / delivered / stop reason / spend. Zero
+examined / qualified / disqualified / delivered / stop reason. Zero
 delivered gets a diagnosis and a concrete next move, never a shrug. Close
 with NEXT STEPS from the tool description \u2014 and STOP; take no further action
 without my say-so.
@@ -1623,7 +1625,7 @@ If \`pull_leads\` itself fails and you have no prior batch, then yes \u2014 retr
 Leadbay has no cancel. Once \`leadbay_enrich_titles\`, \`leadbay_bulk_qualify_leads\`,
 \`leadbay_import_leads\` or \`leadbay_import_and_qualify\` has returned a launched or
 running result, that work is queued on Leadbay and runs to completion, and the
-quota it costs is already committed. A discovery, preview or \`dry_run\` result
+quota it uses is already committed. A discovery, preview or \`dry_run\` result
 launched nothing and is not covered here.
 
 The user cancelling in the chat, a request timeout, or a closed stream stops YOUR
@@ -1631,7 +1633,7 @@ waiting, never the job. \`cancelled: true\` means we stopped watching, not that
 work stopped. What to do next depends on what you are holding:
 
 - **A handle.** Poll the status tool with it, and do not launch the work that
-  handle covers a second time \u2014 that spends the quota again on the same rows.
+  handle covers a second time \u2014 that uses the quota again on the same rows.
   \`leadbay_import_status\` takes \`importIds\`, so pass the values of \`import_ids\`
   under that name. A qualification started by \`leadbay_import_and_qualify\` has no
   notification of its own: resume it with
@@ -1645,7 +1647,7 @@ work stopped. What to do next depends on what you are holding:
   finished. Calling the same tool again with the same arguments will usually hand
   back the job already launched rather than starting a second one, but that guard
   is in-memory, five minutes, and per process, so it is best-effort \u2014 say what you
-  are about to re-run before you spend the user's quota on it.
+  are about to re-run before you use the user's quota on it.
 
 
 
@@ -1788,7 +1790,7 @@ All \`Calls\` below are agent-callable \`leadbay_*\` tools (never an MCP prompt
 |---------------------------------------------------------------------|--------------------------------------------------------|--------------------------------------------------------------|
 | Fresh discovery batch waiting / user wants new leads                | "See today's best new leads"                           | leadbay_pull_leads(lensId = pinned)                          |
 | Follow-ups due / known leads to re-engage                           | "Show follow-ups due now"                              | leadbay_pull_followups                                       |
-| Quota/credit read shows low or exhausted balance                    | "Review what's eating your quota"                      | leadbay_account_status (deeper read)                         |
+| Quota/credit read shows exhausted balance                           | "Review what's eating your quota"                      | leadbay_account_status (deeper read)                         |
 | Auth/connection blocker (e.g. 401 / AUTH_EXPIRED on a read)         | "Reconnect Leadbay to unblock actions"                 | (guide the user to re-authenticate \u2014 no tool call)           |
 | Lens audience looks mismatched (batch is off-ICP)                   | "Adjust the lens audience to match your ICP"           | ASK first \u2014 collect the target sectors / sizes / exclusions, THEN leadbay_adjust_audience(...) with those params. NEVER call it with no args (an empty call writes the current filter / may clone the default lens \u2014 a no-op or unwanted change). |
 | Status is healthy and nothing is pending                            | propose nothing \u2014 the overview is a complete answer    | \u2014                                                            |
@@ -2336,7 +2338,7 @@ If \`pull_leads\` itself fails and you have no prior batch, then yes \u2014 retr
 Leadbay has no cancel. Once \`leadbay_enrich_titles\`, \`leadbay_bulk_qualify_leads\`,
 \`leadbay_import_leads\` or \`leadbay_import_and_qualify\` has returned a launched or
 running result, that work is queued on Leadbay and runs to completion, and the
-quota it costs is already committed. A discovery, preview or \`dry_run\` result
+quota it uses is already committed. A discovery, preview or \`dry_run\` result
 launched nothing and is not covered here.
 
 The user cancelling in the chat, a request timeout, or a closed stream stops YOUR
@@ -2344,7 +2346,7 @@ waiting, never the job. \`cancelled: true\` means we stopped watching, not that
 work stopped. What to do next depends on what you are holding:
 
 - **A handle.** Poll the status tool with it, and do not launch the work that
-  handle covers a second time \u2014 that spends the quota again on the same rows.
+  handle covers a second time \u2014 that uses the quota again on the same rows.
   \`leadbay_import_status\` takes \`importIds\`, so pass the values of \`import_ids\`
   under that name. A qualification started by \`leadbay_import_and_qualify\` has no
   notification of its own: resume it with
@@ -2358,7 +2360,7 @@ work stopped. What to do next depends on what you are holding:
   finished. Calling the same tool again with the same arguments will usually hand
   back the job already launched rather than starting a second one, but that guard
   is in-memory, five minutes, and per process, so it is best-effort \u2014 say what you
-  are about to re-run before you spend the user's quota on it.
+  are about to re-run before you use the user's quota on it.
 
 
 
@@ -2579,11 +2581,11 @@ Cash-to-capture is not available: it needs \`ca12\` from my invoicing system, wh
 
 # PHASE 5 \u2014 CONTACTS (consent-gated)
 
-Each card needs a reachable decision-maker. \`leadbay_enrich_titles({leadIds, lensId})\` in discovery mode first \u2014 that reveals what's enrichable and spends nothing. Render whatever contact detail is already on the record; many accounts already carry a named contact.
+Each card needs a reachable decision-maker. \`leadbay_enrich_titles({leadIds, lensId})\` in discovery mode first \u2014 that reveals what's enrichable and uses no quota. Render whatever contact detail is already on the record; many accounts already carry a named contact.
 
-**Do NOT stop and wait for enrichment consent before delivering.** Asking for a plan is not authorization to spend quota on {{arg:count_or_default}} accounts \u2014 but neither is it a reason to end the turn on a spending question with no plan attached. Ship the ranked plan (Phase 6), then **offer** the paid reveal alongside it. The discovery call returned no \`titles\`, so it only told you what's *available* \u2014 **the offer must therefore carry the titles you propose to enrich AND the channels**, not just a volume: "enrich N contacts at these titles (\`<the titles you picked from available_titles / title_suggestions>\`), email only / email + phone \u2014 reveals consume quota". A bare "yes" to a volume-only question is not a mandate to pick titles yourself, and re-running discovery instead of launching wastes a turn.
+**Do NOT stop and wait for enrichment consent before delivering.** Asking for a plan is not authorization to use quota on {{arg:count_or_default}} accounts \u2014 but neither is it a reason to end the turn on a consent question with no plan attached. Ship the ranked plan (Phase 6), then **offer** the reveal alongside it. The discovery call returned no \`titles\`, so it only told you what's *available* \u2014 **the offer must therefore carry the titles you propose to enrich AND the channels**, not just a volume: "enrich N contacts at these titles (\`<the titles you picked from available_titles / title_suggestions>\`), email only / email + phone \u2014 reveals consume quota". A bare "yes" to a volume-only question is not a mandate to pick titles yourself, and re-running discovery instead of launching wastes a turn.
 
-\u26A0 **Do NOT quote a cost or a credits figure.** The per-reveal rate is backend-side and enrichment is gated by quota, not a credit balance; \`credits_remaining\` is advisory context only. A spend number invented to make the offer concrete is the same failure as an invented euro on a card.
+\u26A0 **Do NOT quote a cost or a credits figure.** The per-reveal rate is backend-side and enrichment is gated by quota, not a credit balance; \`credits_remaining\` is advisory context only. A usage number invented to make the offer concrete is the same failure as an invented euro on a card.
 
 On an explicit yes, launch with the agreed \`titles\` + channels, then poll \`leadbay_bulk_enrich_status\` until done and **keep the \`notification_id\` handles** for the deck.
 
@@ -2731,9 +2733,9 @@ ChatGPT exposes the same routing pattern via \`_meta.openai/outputTemplate\`. We
 - One short intro sentence in chat is enough \u2014 "Here are your 5 NYC follow-ups." Then route into the widget.
 
 
-\u26A0 **The deck's contact layer depends on what actually happened in Phase 5.** Bind a \`leadbay_bulk_enrich_status\` resource ONLY if a paid reveal was launched and you hold a \`notification_id\`. If the user accepted the deck but not the reveal, render the contacts already on record and carry the paid-reveal offer inside the deck \u2014 never wire a status resource with no handle (it renders permanently empty) and never launch enrichment from the deck to manufacture one.
+\u26A0 **The deck's contact layer depends on what actually happened in Phase 5.** Bind a \`leadbay_bulk_enrich_status\` resource ONLY if a reveal was launched and you hold a \`notification_id\`. If the user accepted the deck but not the reveal, render the contacts already on record and carry the reveal offer inside the deck \u2014 never wire a status resource with no handle (it renders permanently empty) and never launch enrichment from the deck to manufacture one.
 
-On acceptance, call \`leadbay_artifact_kit\`, read its \`usage_guide\` before writing any code, and build a single-file deck. Wire the live layer from the handles you kept: a poll-until-done resource per \`notification_id\` for the qualification pills, and one over \`leadbay_bulk_enrich_status\` for the contacts. \u26A0 **If enrichment already ran this session, bind the existing \`notification_id\` \u2014 re-launching enrichment from the deck double-spends my quota.** Per-card notes and outcomes go through the pre-wired note/outreach view-models (they carry the required verification and \`_triggered_by\` fields; hand-rolling those is where it breaks). Keep the checklists in local storage, and always wire a Refresh \u2014 auto-poll is host-dependent. List every tool the deck calls in its \`mcp_tools\`, and render the bridge-unavailable branch, or the pills silently show empty.
+On acceptance, call \`leadbay_artifact_kit\`, read its \`usage_guide\` before writing any code, and build a single-file deck. Wire the live layer from the handles you kept: a poll-until-done resource per \`notification_id\` for the qualification pills, and one over \`leadbay_bulk_enrich_status\` for the contacts. \u26A0 **If enrichment already ran this session, bind the existing \`notification_id\` \u2014 re-launching enrichment from the deck uses my quota twice.** Per-card notes and outcomes go through the pre-wired note/outreach view-models (they carry the required verification and \`_triggered_by\` fields; hand-rolling those is where it breaks). Keep the checklists in local storage, and always wire a Refresh \u2014 auto-poll is host-dependent. List every tool the deck calls in its \`mcp_tools\`, and render the bridge-unavailable branch, or the pills silently show empty.
 
 # Iron laws
 
@@ -2743,7 +2745,7 @@ On acceptance, call \`leadbay_artifact_kit\`, read its \`usage_guide\` before wr
 - **Deliver first, ask alongside.** Do not end a turn without a ranked list of real accounts. The benchmark, the Tier-1 threshold, the territory, a missing lens, a short question set and an unanswered enrichment offer are all NON-blocking \u2014 carry them next to the plan. Only an unresolvable identity mismatch (whose plan is this?) may stop delivery.
 - **One motif per account, from the closed set of six**, with its deciding evidence stated.
 - **The org's real qualification questions**, read from Leadbay \u2014 never invented.
-- **Consent before any paid enrichment**, and never re-launch a bulk that already exists.
+- **Consent before any enrichment**, and never re-launch a bulk that already exists.
 - **Offer the deck; don't force it.** The chat answer must stand alone as useful.
 - Carry the captured \`lensId\` on the calls whose schema **accepts** it (\`leadbay_pull_leads\`, \`leadbay_bulk_qualify_leads\`, \`leadbay_enrich_titles\`). Do NOT add it to \`leadbay_pull_followups\`, \`leadbay_scan_portfolio_signals\`, \`leadbay_qualify_status\` or \`leadbay_bulk_enrich_status\` \u2014 they declare no such argument and reject unknown properties.
 - Building a plan is not outreaching \u2014 do not send anything and do not call \`leadbay_report_outreach\`.
@@ -2795,7 +2797,7 @@ Call \`leadbay_campaign_call_sheet({campaign_id})\`. The response carries \`summ
 
 - "\u{1F4DE} Start calling now" \u2014 IF \`ready_for_calling\`. Top option when phones are there.
 - "\u2709 Email session instead" \u2014 IF \`ready_for_emailing\` AND \`email_ratio > phone_ratio\`. Don't surface this when calling is more obvious.
-- "\u{1F527} Enrich titles first" \u2014 IF \`needs_enrichment\`. Top option when most leads have no contacts. Phrase as "<N> leads have no reachable contact yet \u2014 enrich titles before we start?" so the user understands the cost.
+- "\u{1F527} Enrich titles first" \u2014 IF \`needs_enrichment\`. Top option when most leads have no contacts. Phrase as "<N> leads have no reachable contact yet \u2014 enrich titles before we start?" so the user knows it uses quota.
 - "\u{1F5FA} View on a map" \u2014 IF \`travel_friendly\` **AND** the user hasn't previously signaled disinterest in maps (check your conversation memory; if you've seen the user dismiss map renders before in this session or saved a "no maps" preference, drop this option).
 
 If the MCP prompt argument \`mode\` was actually supplied, skip the proposal and jump to the matching mode below. If \`mode\` was omitted, do not treat \`call_sheet\` as implicit user consent \u2014 propose first.
@@ -2853,7 +2855,7 @@ Optional: offer to review the \`leadbay_campaign_progression\` for the same camp
 - If the user dictates an outcome that doesn't cleanly map to one of the four epilogue values, ASK ONCE before guessing.
 `;
 var PROMPT_META = {
-  leadbay_build_campaign: { "name": "leadbay_build_campaign", "short_description": 'Build a sales campaign from scratch, autonomously, to a target size:\ndiscover on the lens, qualify, and enrich the buyer titles until `count`\nleads each have a reachable target-title contact \u2014 no pauses, no confirm\ngates. Saves via `leadbay_create_campaign` and renders a one-tap\ncall/email view via `leadbay_campaign_call_sheet`. Trigger on "build me a\ncampaign", "build N leads", "create a campaign from scratch". Work an\nexisting one with `leadbay_work_campaign`.\n', "arguments": [{ "name": "audience", "description": "Optional: a fresh audience to target (e.g. 'dental clinics in Texas'). Omit to build from your ACTIVE lens \u2014 the default.", "required": false }, { "name": "campaign_name", "description": "Optional: a name for the campaign. Omit and one is derived from the lens/audience + date (or the backend AI-names it).", "required": false }, { "name": "count", "description": "Optional: how many fully-actionable leads to build (default 20). The loop keeps discovering, qualifying and enriching until this many in-ICP leads each have a reachable target-title contact \u2014 or the lens is exhausted. Higher counts take longer and consume more quota.", "required": false }, { "name": "job_titles", "description": "Optional: the exact buyer job titles to enrich, comma-separated (e.g. 'VP Sales, Head of Growth, Director of Business Development'). Omit and the buyer persona is derived from what you sell. A lead only counts toward the target when it has a reachable contact matching one of these titles.", "required": false }], "expected_calls": ["leadbay_account_status", "leadbay_pull_leads", "leadbay_bulk_qualify_leads", "leadbay_qualify_status", "leadbay_recall_ordered_titles", "leadbay_enrich_titles", "leadbay_bulk_enrich_status", "leadbay_create_campaign", "leadbay_add_leads_to_campaign", "leadbay_campaign_call_sheet", "leadbay_campaign_progression", "leadbay_new_lens", "leadbay_adjust_audience"], "failure_modes": ["Pauses to confirm before enriching (or asks 'enrich these N now?' via ask_user_input_v0) \u2014 this prompt runs to goal with NO confirm gate; asking for the campaign IS the authorization. Never stop for a spend confirmation.", "Stops to confirm a lens switch when the user named a fresh audience \u2014 naming the audience IS the authorization; switch, state which lens in one line, and don't ask.", "Pauses at any point to ask the user to choose, confirm, or hand off \u2014 the only acceptable stops are lens exhaustion (can't supply the count) or a backend 429 (quota out). Anything else is purpose drift.", "Stops at fewer than the target `count` of actionable leads without looping back to pull / qualify / enrich more \u2014 must run to the target, or honestly report the lens is exhausted and offer to widen.", "Counts a lead toward `count` before its target-title contact actually landed (email/phone present) \u2014 an empty enrichment doesn't count; swap and re-enrich until the cohort is genuinely `count` deep.", "Enriches by seniority instead of by buyer persona \u2014 picks COO / Director of Operations / Mgr of Logistics / CFO / CTO because they sound senior, when the user sells a SALES tool whose buyer is the revenue org (VP/Head/Director of Sales, BD, growth, marketing). Operations people never buy a sales tool; this hands the salesperson a useless list.", "When no titles are given, fails to derive the user's buyer persona from their product/ICP before choosing titles \u2014 jumps straight to generic exec titles instead of working out who buys what THIS user sells. (When titles ARE given, use them verbatim \u2014 don't substitute 'more senior' ones.)", "Blindly repeats leadbay_recall_ordered_titles / discovery suggestions even when they are off-persona (e.g. operations roles a prior session wrongly enriched) \u2014 recall is a filtered input, not the answer.", "Poor coverage \u2014 leaves picked leads with no target-title contact (or 0 enrichments on some leads) and ships them anyway, so the salesperson opens the campaign to half-empty rows. Swap them out and refill to the count instead.", "Creates the campaign before qualifying / picking \u2014 seeds a campaign with unvetted leads. Qualify and lock the buyer-covered cohort FIRST, then leadbay_create_campaign.", "Ends at 'campaign created' without rendering the leadbay_campaign_call_sheet view \u2014 the ready-to-work view IS the deliverable; stopping short is purpose drift.", "Runs the calling / outcome / follow-up loop, or calls leadbay_work_campaign itself, instead of stopping at the call sheet \u2014 this prompt BUILDS and stops; work_campaign is a separate session the user starts later.", "Auto-sends outreach or calls leadbay_report_outreach \u2014 building a campaign is not outreaching. No send, no log.", "Re-pulls leadbay_pull_leads without the captured lensId \u2014 a mid-session lens shift discards the cohort being built.", "Renders the picked leads or the call sheet as prose instead of the canonical per-tool RENDERING layout."] },
+  leadbay_build_campaign: { "name": "leadbay_build_campaign", "short_description": 'Build a sales campaign from scratch, autonomously, to a target size:\ndiscover on the lens, qualify, and enrich the buyer titles until `count`\nleads each have a reachable target-title contact \u2014 no pauses, no confirm\ngates. Saves via `leadbay_create_campaign` and renders a one-tap\ncall/email view via `leadbay_campaign_call_sheet`. Trigger on "build me a\ncampaign", "build N leads", "create a campaign from scratch". Work an\nexisting one with `leadbay_work_campaign`.\n', "arguments": [{ "name": "audience", "description": "Optional: a fresh audience to target (e.g. 'dental clinics in Texas'). Omit to build from your ACTIVE lens \u2014 the default.", "required": false }, { "name": "campaign_name", "description": "Optional: a name for the campaign. Omit and one is derived from the lens/audience + date (or the backend AI-names it).", "required": false }, { "name": "count", "description": "Optional: how many fully-actionable leads to build (default 20). The loop keeps discovering, qualifying and enriching until this many in-ICP leads each have a reachable target-title contact \u2014 or the lens is exhausted. Higher counts take longer and consume more quota.", "required": false }, { "name": "job_titles", "description": "Optional: the exact buyer job titles to enrich, comma-separated (e.g. 'VP Sales, Head of Growth, Director of Business Development'). Omit and the buyer persona is derived from what you sell. A lead only counts toward the target when it has a reachable contact matching one of these titles.", "required": false }], "expected_calls": ["leadbay_account_status", "leadbay_pull_leads", "leadbay_bulk_qualify_leads", "leadbay_qualify_status", "leadbay_recall_ordered_titles", "leadbay_enrich_titles", "leadbay_bulk_enrich_status", "leadbay_create_campaign", "leadbay_add_leads_to_campaign", "leadbay_campaign_call_sheet", "leadbay_campaign_progression", "leadbay_new_lens", "leadbay_adjust_audience"], "failure_modes": ["Pauses to confirm before enriching (or asks 'enrich these N now?' via ask_user_input_v0) \u2014 this prompt runs to goal with NO confirm gate; asking for the campaign IS the authorization. Never stop for an enrichment confirmation.", "Stops to confirm a lens switch when the user named a fresh audience \u2014 naming the audience IS the authorization; switch, state which lens in one line, and don't ask.", "Pauses at any point to ask the user to choose, confirm, or hand off \u2014 the only acceptable stops are lens exhaustion (can't supply the count) or a backend 429 (quota out). Anything else is purpose drift.", "Stops at fewer than the target `count` of actionable leads without looping back to pull / qualify / enrich more \u2014 must run to the target, or honestly report the lens is exhausted and offer to widen.", "Counts a lead toward `count` before its target-title contact actually landed (email/phone present) \u2014 an empty enrichment doesn't count; swap and re-enrich until the cohort is genuinely `count` deep.", "Enriches by seniority instead of by buyer persona \u2014 picks COO / Director of Operations / Mgr of Logistics / CFO / CTO because they sound senior, when the user sells a SALES tool whose buyer is the revenue org (VP/Head/Director of Sales, BD, growth, marketing). Operations people never buy a sales tool; this hands the salesperson a useless list.", "When no titles are given, fails to derive the user's buyer persona from their product/ICP before choosing titles \u2014 jumps straight to generic exec titles instead of working out who buys what THIS user sells. (When titles ARE given, use them verbatim \u2014 don't substitute 'more senior' ones.)", "Blindly repeats leadbay_recall_ordered_titles / discovery suggestions even when they are off-persona (e.g. operations roles a prior session wrongly enriched) \u2014 recall is a filtered input, not the answer.", "Poor coverage \u2014 leaves picked leads with no target-title contact (or 0 enrichments on some leads) and ships them anyway, so the salesperson opens the campaign to half-empty rows. Swap them out and refill to the count instead.", "Creates the campaign before qualifying / picking \u2014 seeds a campaign with unvetted leads. Qualify and lock the buyer-covered cohort FIRST, then leadbay_create_campaign.", "Ends at 'campaign created' without rendering the leadbay_campaign_call_sheet view \u2014 the ready-to-work view IS the deliverable; stopping short is purpose drift.", "Runs the calling / outcome / follow-up loop, or calls leadbay_work_campaign itself, instead of stopping at the call sheet \u2014 this prompt BUILDS and stops; work_campaign is a separate session the user starts later.", "Auto-sends outreach or calls leadbay_report_outreach \u2014 building a campaign is not outreaching. No send, no log.", "Re-pulls leadbay_pull_leads without the captured lensId \u2014 a mid-session lens shift discards the cohort being built.", "Renders the picked leads or the call sheet as prose instead of the canonical per-tool RENDERING layout."] },
   leadbay_daily_check_in: { "name": "leadbay_daily_check_in", "short_description": 'Morning DISCOVERY workflow \u2014 new leads from the lens wishlist. Trigger\non "show me leads", "what\'s new today", "let\'s prospect", "run my check-in",\n"my morning check-in", "I do this every day", "every morning". Recurrence\nlanguage always means this prompt. Do NOT trigger on follow-up phrasings\n("follow up", "before my trip") \u2014 those go to `leadbay_followup_check_in`.\n', "arguments": [], "expected_calls": ["leadbay_account_status", "leadbay_pull_leads", "leadbay_research_lead_by_id", "leadbay_bulk_qualify_leads", "leadbay_enrich_contacts"], "failure_modes": ["Calls leadbay_report_outreach without explicit user authorization", "Surfaces fewer than 10 leads when more are available, or fails to top up via leadbay_qualify_top_n when the batch is short", `Replaces the canonical pull_leads table layout with prose per row (the per-tool RENDERING block is the structural contract; "Today's nudges" goes above it, not in place of it)`, "Skips the nudge paragraph entirely \u2014 the table alone is fine but adding the nudge is the value-add", `Skips deep research on promising leads (Phase 4) \u2014 the agent must call leadbay_research_lead_by_id on each when the user's intent is to research specific leads; Phase 4 is intentionally skipped for batch-view requests ("show me today's leads", "run my morning check-in") per the Phase 4 skip gate`, "Triggers contact enrichment without asking the user first (it consumes quota)", "Skips the STOP byproduct and proposes next actions on its own", 'Fires 10 parallel leadbay_research_lead_by_id calls and treats "stream closed" errors as terminal \u2014 must serialize and retry singletons', "Re-pulls leadbay_pull_leads without passing the captured lensId, allowing a backend lens shift to discard the Phase 2 batch", 'Treats a "Request timed out" from leadbay_bulk_qualify_leads as terminal instead of retrying with wait_for_completion:false + qualify_status polling', 'Triggers on a follow-up query (e.g., "leads I should follow up with") that should have routed to `leadbay_followup_check_in` \u2014 the two entry points are different data sources (Discover wishlist vs Monitor view) per \xA71.6'] },
   leadbay_extend_my_lens: { "name": "leadbay_extend_my_lens", "short_description": "Add more leads to the current lens on demand \u2014 for users whose appetite\nexceeds the standard daily fill. The agent picks seeds silently from\nwhat's already on the lens, fires the extra refill, and surfaces the\nqueue confirmation. The user never reviews the seed list.\n", "arguments": [{ "name": "extra_count", "description": "How many extra leads to add. Optional. Omit to use the backend default.", "required": false }], "expected_calls": ["leadbay_account_status", "leadbay_seed_candidates", "leadbay_extend_lens", "leadbay_pull_leads"], "failure_modes": ["Surfaces the seed candidate list to the user instead of picking silently \u2014 the user asked for MORE LEADS, not a candidate review meeting", "Skips the seeded path and calls `leadbay_extend_lens` with no `seed_lead_ids`, losing the bias signal the recommender needs", "On 429, silently retries instead of surfacing the three options (smaller / wait / upgrade) via your host's choice widget (`ask_user_input_v0` or `AskUserQuestion`)", "Forgets to pre-check `LENS_EXTRA_REFILL` quota in `leadbay_account_status` and burns a wasted API call", "Skips the post-queue pull-leads suggestion, so the user doesn't see what just got added"] },
   leadbay_followup_check_in: { "name": "leadbay_followup_check_in", "short_description": 'Follow-up check-in: surface KNOWN leads from the Monitor view needing\nre-engagement. Trigger on "follow up", "already known leads", "what\'s\noverdue", "before my trip", "who should I re-engage". Do NOT trigger on\n"show me today\'s leads", "my morning check-in", "run my check-in",\n"I do this every day", "every morning" \u2014 those go to\n`leadbay_daily_check_in`.\n', "arguments": [], "expected_calls": ["leadbay_pull_followups", "leadbay_research_lead_by_id", "leadbay_prepare_outreach"], "failure_modes": ["Calls leadbay_pull_leads (the Discover entry point) instead of leadbay_pull_followups \u2014 these are different data sources; the Discover queue does NOT contain Monitor's known-but-cold pipeline", 'Iterates pages of leadbay_pull_leads filtering by engagement_count to "fake" a follow-up view (a real bug observed in 0.9.0 \u2014 the right move is to call pull_followups directly)', "Replaces the canonical pull_followups table layout with prose per row (the per-tool RENDERING block is the structural contract; commentary belongs above or below)", 'Skips the cross-mode pivot offer at the end ("Want to see NEW leads from your wishlist instead?" routes to leadbay_pull_leads)'] },
@@ -2863,7 +2865,7 @@ reveal who to send it to. Use when the user is new or asks to be SHOWN how
 Leadbay works ("walk me through Leadbay", "I'm new", "how do I use this",
 "give me a tour"). Don't use it for orientation prose with no clicking \u2014
 that's leadbay_prospecting_overview.
-`, "arguments": [], "expected_calls": ["leadbay_account_status", "leadbay_pull_leads", "leadbay_prepare_outreach", "leadbay_enrich_titles", "leadbay_bulk_enrich_status"], "failure_modes": ['Presents a gate as prose ("let me know if you want me to pull your leads") instead of CALLING the host choice widget \u2014 the click IS the lesson, and prose turns the walkthrough into a lecture', "Runs a step's tool WITHOUT firing that step's widget first and waiting for the click \u2014 the walkthrough becomes an automated demo the user only watches, which is the exact opposite of learning by doing", "Fires the widget without the EXPLAIN beat, so the user gets an unexplained button and learns nothing about what a lens or an enrichment actually is", 'Answers gate 1 with a bare "you\'re connected as X at Y" when the quota IS readable \u2014 the user clicked a button labelled `check my account status`, so the quota windows (Daily/Weekly/Monthly gauges, % used, $ spent, resets) ARE the answer, not an optional extra', 'Renders quota as raw "credits" instead of the web app\'s percentage + dollar-spend gauges, or dumps raw `resource_type` strings the user has never seen', "Opens with a wall of text \u2014 previewing all four steps, explaining lenses up front, or writing several paragraphs before the first widget. The opening is TWO lines then the button; a first-run user wants to see it work, not read a syllabus", `Ends the first message without firing gate 1's widget, leaving the user to reply "ok" before anything happens`, "Rewrites the gate's own `next_steps` payload (its `question`, `label` or `description`) instead of mapping it into the widget verbatim, or merges two gates into a single multi-option widget", 'Fires a THIRD option, or turns the exit into an alternative route ("show me my lenses instead") \u2014 each gate carries exactly one forward action plus the `I\'m done for now` exit, never a menu of paths', `Fires a single-option widget \u2014 the host requires 2\u20134 options, so a lone option is rejected or silently degrades to prose ("say the word and I'll check it"), which is the exact defect this rule exists to prevent`, 'Launches the PAID reveal at gate 4 BEFORE the user has picked leads and confirmed \u2014 beat 1 must be the free `mode:"discover"` preview (no `titles`, no `confirm`, no `email`, no `phone`); the gate click bought the free look, not the reveal, and silence is never consent', "Stops at the free preview after the user DID pick leads and confirm \u2014 they asked for real contact details, so the second call must actually run with `confirm:true` and the chosen titles", "Reports the enrichment without polling `leadbay_bulk_enrich_status` to completion, so it claims contacts it never actually saw resolve", `Reveals contacts and never says what it cost \u2014 the user just spent credits and deserves the one-line "N contacts = N credits", which is also what makes gate 1's quota numbers concrete`, 'Reports "no leads" on an empty batch while `computing_wishlist` / `computing_scores` is true \u2014 the lens is still building; render the tool\'s own two-option warm-up widget verbatim and pause', "Rewords, reorders or prose-ifies the `next_steps` payload from `leadbay_pull_leads` instead of mapping `options[]` into the widget verbatim", "Runs all four steps in one turn without waiting for the user's click between gates \u2014 the walkthrough is a sequence of gates, not a script to recite", "Skips `leadbay_pull_leads` and jumps straight to enrichment, leaving gate 4 with no `leadIds` to scope", "Passes a singular `leadId` to `leadbay_enrich_titles` on the confirmed reveal \u2014 that key does not exist on this tool, so it is dropped and the paid call falls back to the whole default wishlist selection, charging for far more than the one lead the user agreed to. it is always the `leadIds` ARRAY, even for a single lead", "Drops the pinned `lens.id` between gates, so gate 4 enriches against a different lens than the one the user just saw", "Ends the completed walkthrough without the `keep_going` cheat-sheet \u2014 the buttons disappear with the tour, so a user who was never told what to TYPE learned to click a tutorial and nothing about using Leadbay tomorrow", "Invents phrases for the cheat-sheet, or rewords them into something that sounds nicer but doesn't match the tool's real triggers \u2014 teaching a phrase that doesn't route is worse than teaching none", "SENDS the gate 3 draft, or offers to send it \u2014 the walkthrough drafts and stops there; the email is the user's to judge, and nothing leaves the chat", "Passes `enrich:true` to `leadbay_prepare_outreach` at gate 3 \u2014 that launches a PAID contact reveal off the back of a DRAFT click, spending credits the user never agreed to", "Invents a contact NAME for the gate 3 draft \u2014 `recommended_contact` still has null email/name at that point, so the draft is addressed to the job TITLE; a fabricated name is the one thing that makes the whole draft untrustworthy", "Treats the null email at gate 3 as a failure \u2014 apologising for it, retrying, or calling another tool to fill it in. It is the setup for gate 4 \u2014 an email written, nobody to send it to yet", "Pastes the drafted email into chat prose alongside `message_compose_v1` instead of letting the composer BE the answer", "Enriches leads other than the one it drafted for at gate 3 \u2014 gate 4 reveals the person that email is going to, so it is scoped to that ONE lead, one contact, one credit", "Renders the cheat-sheet on the exit and stops there, dropping the 1:1 offer \u2014 the observed failure is that the agent feels finished once the table is on screen, so the user who just stepped out never hears about the help that would bring them back. ENDING B is not complete without the offer, and the offer goes LAST", "Treats the exit click as ENDING C (typed off-script) and closes in silence, or treats a typed request as ENDING B and buries their real answer under a cheat-sheet and a booking link", "Turns the exit offer into a pitch \u2014 several sentences, a re-opened gate, or an argument for finishing the tour. They said they were done; it is one line and a link", "Fires the 1:1 offer mid-tour, or at a user who left by TYPING a different request \u2014 a booking link on top of their real question is an interruption, not an offer", "Runs the four gates at a user whose actual problem is SETUP \u2014 the connector isn't installed, they can't sign in, or their Leadbay tools aren't appearing. The tour assumes a working connection and cannot fix any of it; the setup guide can", "Pastes the setup-guide link mid-tour, between gates, instead of once at the closing \u2014 a link in the middle of the walkthrough invites the user to leave the thing they're doing"] },
+`, "arguments": [], "expected_calls": ["leadbay_account_status", "leadbay_pull_leads", "leadbay_prepare_outreach", "leadbay_enrich_titles", "leadbay_bulk_enrich_status"], "failure_modes": ['Presents a gate as prose ("let me know if you want me to pull your leads") instead of CALLING the host choice widget \u2014 the click IS the lesson, and prose turns the walkthrough into a lecture', "Runs a step's tool WITHOUT firing that step's widget first and waiting for the click \u2014 the walkthrough becomes an automated demo the user only watches, which is the exact opposite of learning by doing", "Fires the widget without the EXPLAIN beat, so the user gets an unexplained button and learns nothing about what a lens or an enrichment actually is", 'Answers gate 1 with a bare "you\'re connected as X at Y" when the quota IS readable \u2014 the user clicked a button labelled `check my account status`, so the quota windows (Daily/Weekly/Monthly gauges, % used, $ spent, resets) ARE the answer, not an optional extra', 'Renders quota as raw "credits" instead of the web app\'s percentage + dollar-spend gauges, or dumps raw `resource_type` strings the user has never seen', "Opens with a wall of text \u2014 previewing all four steps, explaining lenses up front, or writing several paragraphs before the first widget. The opening is TWO lines then the button; a first-run user wants to see it work, not read a syllabus", `Ends the first message without firing gate 1's widget, leaving the user to reply "ok" before anything happens`, "Rewrites the gate's own `next_steps` payload (its `question`, `label` or `description`) instead of mapping it into the widget verbatim, or merges two gates into a single multi-option widget", 'Fires a THIRD option, or turns the exit into an alternative route ("show me my lenses instead") \u2014 each gate carries exactly one forward action plus the `I\'m done for now` exit, never a menu of paths', `Fires a single-option widget \u2014 the host requires 2\u20134 options, so a lone option is rejected or silently degrades to prose ("say the word and I'll check it"), which is the exact defect this rule exists to prevent`, 'Launches the contact reveal at gate 4 BEFORE the user has picked leads and confirmed \u2014 beat 1 must be the free `mode:"discover"` preview (no `titles`, no `confirm`, no `email`, no `phone`); the gate click bought the free look, not the reveal, and silence is never consent', "Stops at the free preview after the user DID pick leads and confirm \u2014 they asked for real contact details, so the second call must actually run with `confirm:true` and the chosen titles", "Reports the enrichment without polling `leadbay_bulk_enrich_status` to completion, so it claims contacts it never actually saw resolve", "Reveals contacts and never says the reveal used a little of their plan's quota \u2014 that one line, with no amount and no price, is what makes gate 1's quota numbers concrete", 'Reports "no leads" on an empty batch while `computing_wishlist` / `computing_scores` is true \u2014 the lens is still building; render the tool\'s own two-option warm-up widget verbatim and pause', "Rewords, reorders or prose-ifies the `next_steps` payload from `leadbay_pull_leads` instead of mapping `options[]` into the widget verbatim", "Runs all four steps in one turn without waiting for the user's click between gates \u2014 the walkthrough is a sequence of gates, not a script to recite", "Skips `leadbay_pull_leads` and jumps straight to enrichment, leaving gate 4 with no `leadIds` to scope", "Passes a singular `leadId` to `leadbay_enrich_titles` on the confirmed reveal \u2014 that key does not exist on this tool, so it is dropped and the reveal falls back to the whole default wishlist selection, revealing far more than the one lead the user agreed to. it is always the `leadIds` ARRAY, even for a single lead", "Drops the pinned `lens.id` between gates, so gate 4 enriches against a different lens than the one the user just saw", "Ends the completed walkthrough without the `keep_going` cheat-sheet \u2014 the buttons disappear with the tour, so a user who was never told what to TYPE learned to click a tutorial and nothing about using Leadbay tomorrow", "Invents phrases for the cheat-sheet, or rewords them into something that sounds nicer but doesn't match the tool's real triggers \u2014 teaching a phrase that doesn't route is worse than teaching none", "SENDS the gate 3 draft, or offers to send it \u2014 the walkthrough drafts and stops there; the email is the user's to judge, and nothing leaves the chat", "Passes `enrich:true` to `leadbay_prepare_outreach` at gate 3 \u2014 that launches a contact reveal off the back of a DRAFT click, using quota the user never agreed to", "Invents a contact NAME for the gate 3 draft \u2014 `recommended_contact` still has null email/name at that point, so the draft is addressed to the job TITLE; a fabricated name is the one thing that makes the whole draft untrustworthy", "Treats the null email at gate 3 as a failure \u2014 apologising for it, retrying, or calling another tool to fill it in. It is the setup for gate 4 \u2014 an email written, nobody to send it to yet", "Pastes the drafted email into chat prose alongside `message_compose_v1` instead of letting the composer BE the answer", "Enriches leads other than the one it drafted for at gate 3 \u2014 gate 4 reveals the person that email is going to, so it is scoped to that ONE lead, one contact", "Renders the cheat-sheet on the exit and stops there, dropping the 1:1 offer \u2014 the observed failure is that the agent feels finished once the table is on screen, so the user who just stepped out never hears about the help that would bring them back. ENDING B is not complete without the offer, and the offer goes LAST", "Treats the exit click as ENDING C (typed off-script) and closes in silence, or treats a typed request as ENDING B and buries their real answer under a cheat-sheet and a booking link", "Turns the exit offer into a pitch \u2014 several sentences, a re-opened gate, or an argument for finishing the tour. They said they were done; it is one line and a link", "Fires the 1:1 offer mid-tour, or at a user who left by TYPING a different request \u2014 a booking link on top of their real question is an interruption, not an offer", "Runs the four gates at a user whose actual problem is SETUP \u2014 the connector isn't installed, they can't sign in, or their Leadbay tools aren't appearing. The tour assumes a working connection and cannot fix any of it; the setup guide can", "Pastes the setup-guide link mid-tour, between gates, instead of once at the closing \u2014 a link in the middle of the walkthrough invites the user to leave the thing they're doing"] },
   leadbay_import_file: { "name": "leadbay_import_file", "short_description": "Import a user-supplied CSV/file into Leadbay through five phases with\nevidence gates \u2014 scan, derive, resolve identities, preserve & commit,\nthen optionally qualify and report. The job is to maximize how many\nrows the Leadbay system actually ingests and matches.\n", "arguments": [{ "name": "file", "description": "Path or user-visible name of the CSV/file to import. If omitted, use the file the user attached or referenced.", "required": false }, { "name": "instruction", "description": 'Additional user goal, e.g. "then qualify the leads", "preserve owner phone as a custom field", or "only import restaurants in Manhattan".', "required": false }], "expected_calls": ["leadbay_resolve_import_rows", "leadbay_list_mappable_fields", "leadbay_create_custom_field", "leadbay_import_leads", "leadbay_import_and_qualify", "leadbay_add_note", "leadbay_import_status"], "failure_modes": ["Picks LEADBAY_ID from score alone, name-only, fuzzy-name-only, root-domain-only, brand-only, postcode-only, or city-only evidence", "Drops meaningful business notes or CRM record links instead of preserving them as custom fields or lead notes", "Treats a consumer mailbox domain (gmail.com, hotmail.com, ...) as the company domain", "Skips deriving company_domain from a business email when no website column exists (this kills match rate)", "Skips the COLUMN PRESERVATION PLAN byproduct before importing", "Skips the DECISION LOG byproduct before writing LEADBAY_ID", "Returns the imported records WITHOUT writing LEADBAY_ID values back into the user's file (leaves the user no audit trail of what matched)", "Fabricates leadIds, contact emails, or mapping IDs not present in the file or a tool response"] },
   leadbay_log_outreach: { "name": "leadbay_log_outreach", "short_description": "Log outreach (an email I sent, a call I made, a meeting I had) on a\nspecific lead. Captures verification so the SDR pipeline trusts the entry.\n", "arguments": [{ "name": "lead_id", "description": "The lead UUID. Get it from leadbay_pull_leads or leadbay_research_lead_by_id.", "required": true }, { "name": "summary", "description": "1-2 sentences describing what I did (e.g. 'Sent intro email to CTO citing recent Hornsea contract').", "required": true }], "expected_calls": ["leadbay_report_outreach"], "failure_modes": ["Calls leadbay_report_outreach without first collecting a verification source", "Fabricates a gmail_message_id or calendar_event_id (the human team treats verification as canonical)", "Records outreach to a different lead_id than the one the user supplied", "Skips the dry_run step when the user is unsure what would be sent"] },
   leadbay_new_leads: { "name": "leadbay_new_leads", "short_description": `Guided net-new lead delivery \u2014 turn a described need ("gyms around Dallas
@@ -2873,7 +2875,7 @@ Trigger when the user DESCRIBES who they want: "get me N companies that
 <profile>", "we're entering <market>". A bare "find me new leads" with no
 profile, and "today's leads", are the daily lens batch \u2014 leadbay_pull_leads.
 "Qualify these companies I have" is leadbay_qualify_leads.
-`, "arguments": [{ "name": "need", "description": "What the user is looking for, in their own words (e.g. '10 gyms around Dallas that would buy modular flooring, with phone numbers'). Optional \u2014 the session starts by asking when absent.", "required": false }], "expected_calls": ["leadbay_find_new_leads", "leadbay_lead_job_status", "leadbay_qualify_leads"], "failure_modes": ["Passes the user's raw sentence as `query` instead of crafting an example_lead description (vendor-vocabulary trap \u2014 measured 0 delivered from a raw query vs on-profile results from a crafted example)", "Invents a distinctive brand name in example_lead.name (pulls matching toward name-lookalikes)", 'Puts event language ("hiring", "expanding", "just raised") into the seed description', "Launches qualify:true or channels without a dry_run quote and the user's explicit go-ahead", "Retries a failed/timed-out submit with a NEW request_id (double-spend) \u2014 the same ask must reuse the same request_id", 'Reports "no results" without narrating the funnel + scope_notes and proposing a concrete fix', "Renders delivered leads as freeform prose instead of the canonical lead-delivery table", "Blends two distinct buyer archetypes into one seed description instead of running one search per archetype", "Passes a country name in filters.locations (silently matches a same-named town \u2014 whole-country intent means OMITTING locations) or a nested employees object instead of the flat employees_min/employees_max", "Renders rows that visibly violate the user's exclusions, or presents a best-fit-under-30 table as an answer instead of flagging weak matches"] },
+`, "arguments": [{ "name": "need", "description": "What the user is looking for, in their own words (e.g. '10 gyms around Dallas that would buy modular flooring, with phone numbers'). Optional \u2014 the session starts by asking when absent.", "required": false }], "expected_calls": ["leadbay_find_new_leads", "leadbay_lead_job_status", "leadbay_qualify_leads"], "failure_modes": ["Passes the user's raw sentence as `query` instead of crafting an example_lead description (vendor-vocabulary trap \u2014 measured 0 delivered from a raw query vs on-profile results from a crafted example)", "Invents a distinctive brand name in example_lead.name (pulls matching toward name-lookalikes)", 'Puts event language ("hiring", "expanding", "just raised") into the seed description', "Launches qualify:true or channels without a dry_run quote and the user's explicit go-ahead", "Retries a failed/timed-out submit with a NEW request_id (double launch) \u2014 the same ask must reuse the same request_id", 'Reports "no results" without narrating the funnel + scope_notes and proposing a concrete fix', "Renders delivered leads as freeform prose instead of the canonical lead-delivery table", "Blends two distinct buyer archetypes into one seed description instead of running one search per archetype", "Passes a country name in filters.locations (silently matches a same-named town \u2014 whole-country intent means OMITTING locations) or a nested employees object instead of the flat employees_min/employees_max", "Renders rows that visibly violate the user's exclusions, or presents a best-fit-under-30 table as an answer instead of flagging weak matches"] },
   leadbay_plan_tour_in_city: { "name": "leadbay_plan_tour_in_city", "short_description": 'Use whenever the user names a city they\'ll be in and asks who to see\n\u2014 "I\'m in SF next Tuesday, who\'s worth meeting?", "I\'m going to Berlin\n\u2014 who should I visit?", "plan my <city> tour". Any in-person/visit\nintent tied to a place routes here, NOT to `leadbay_pull_leads`. It\nsurfaces follow-ups + fresh Discover leads in the city via\n`leadbay_tour_plan`, ALWAYS offers to plot them on a map (rendering it\non yes), then offers outreach drafts + campaign persistence.\n', "arguments": [{ "name": "city", "description": "City or region the user is visiting (e.g. 'Limoges', 'Bay Area'). Used as the geo filter for both Monitor and Discover lookups. A country is not a city: this workspace already covers exactly one country, and a country name here silently fences the tour to a same-named village. Do NOT omit the argument to recover \u2014 a city-less tour returns arbitrary leads from across the whole workspace, which is not an itinerary. Ask which city or region the visit is to.", "required": true }, { "name": "date", "description": "When the visit is (e.g. 'May 24', 'next Thursday'). Surfaced in the outreach drafts as 'I'll be in <city> on <date>'.", "required": false }], "expected_calls": ["leadbay_tour_plan", "leadbay_research_lead_by_id", "leadbay_prepare_outreach", "leadbay_create_campaign"], "failure_modes": ["Calls leadbay_followups_map (Monitor-only) instead of leadbay_tour_plan \u2014 loses the Discover (fresh-lead) half that the user explicitly asked for", "Calls leadbay_pull_leads then drops the geo filter \u2014 returns the lens-wide wishlist instead of city-relevant fresh leads", 'Skips the campaign-persist step ("would you like to save these as a tour?") \u2014 leaves the rep with a one-shot map but no follow-up artifact', "Creates a campaign WITHOUT asking the user first \u2014 the persist step is high-intent; offer it, don't assume", "Fabricates lead_ids when seeding the campaign instead of using the ids returned by tour_plan"] },
   leadbay_prospecting_overview: { "name": "leadbay_prospecting_overview", "short_description": `Orientation for working with Leadbay from any host \u2014 discovery vs.
 follow-up, the outreach loop, outcome recording, imports, pushback /
@@ -2886,7 +2888,7 @@ should I follow up on" to "I'll send via lemlist".
   leadbay_refine_audience: { "name": "leadbay_refine_audience", "short_description": "Refine the kind of leads Leadbay surfaces beyond firmographics, with a\nfree-text instruction. Handles the clarification round-trip if the new\nprompt is ambiguous.\n", "arguments": [{ "name": "instruction", "description": "The refinement (e.g. 'focus on hospitals running their own IT'). Set to plain English.", "required": true }], "expected_calls": ["leadbay_refine_prompt", "leadbay_account_status"], "failure_modes": ["Calls leadbay_answer_clarification on the user's behalf instead of surfacing the clarification verbatim", "Glosses over the clarification options instead of presenting them as offered", "Promises immediate effect when status='applied' actually triggers an async intelligence recompute"] },
   leadbay_research_a_domain: { "name": "leadbay_research_a_domain", "short_description": "Resolve a company by name or domain across the user's visible Discover,\nMonitor, and Activate corpus, then return everything Leadbay knows about it.\n", "arguments": [{ "name": "domain", "description": "Company name or domain (for example 'Acme Corporation' or 'acme.com'). The legacy argument key remains `domain` for client compatibility.", "required": true }], "expected_calls": ["leadbay_research_lead_by_name_fuzzy"], "failure_modes": ["Fabricates qualification answers not present in any tool response", "Calls leadbay_import_and_qualify before searching the existing visible corpus", "Treats the active lens as the entire search universe when the user did not request a lens scope", "Imports a missing company without the user's explicit permission", "Renders the research result as a freeform narrative instead of the canonical research-company-card layout (the card with header score bar, pill row, signal sections, contacts table is the structural contract; commentary belongs ABOVE or BELOW it)"] },
   leadbay_setup_team_prospecting: { "name": "leadbay_setup_team_prospecting", "short_description": "Manager-led prospecting setup: conversationally turn a natural-language\naudience ask into a Leadbay lens, validate the candidate leads, and\npersist them as one or more named campaigns the rep(s) can work\nthrough. Closes #3630 US3 end-to-end (within the current\ncreator-scoped campaign visibility model).\n", "arguments": [{ "name": "audience", "description": "Natural-language audience description (e.g. 'plumbing companies with 10-50 employees in Seine-Maritime'). The lens-creation step (`leadbay_refine_prompt` \u2192 `leadbay_create_lens`) interprets it. A country name is not a scope here \u2014 this workspace already covers exactly one country, so drop it and keep the rest of the description; a DIFFERENT country cannot be targeted at all.", "required": true }, { "name": "rep_split", "description": "Optional: how to split the validated leads into per-rep campaigns. Free text \u2014 e.g. 'split by city' or 'one campaign per rep: John gets Tulsa, Sarah gets OKC'. Splitting by country is not a split \u2014 the workspace is single-country.", "required": false }], "expected_calls": ["leadbay_refine_prompt", "leadbay_create_lens", "leadbay_promote_lens", "leadbay_pull_leads", "leadbay_research_lead_by_id", "leadbay_create_campaign", "leadbay_add_leads_to_campaign"], "failure_modes": ["Skips the validation step \u2014 creates a campaign of unvetted leads from a freshly-created lens without giving the manager a chance to drop weak fits", "Creates ONE campaign for all reps without asking about the split \u2014 the user explicitly mentioned per-rep distribution and the prompt should honor it", "Pretends the backend supports cross-user assignment \u2014 campaigns are owned by the caller (creator-scoped). Surface this honestly instead of fabricating an assignment model", "Asks ALL clarifying questions inline before tool calls \u2014 instead, run the lens refinement loop with `leadbay_refine_prompt` which handles the clarification protocol natively"] },
-  leadbay_top_accounts_to_activate: { "name": "leadbay_top_accounts_to_activate", "short_description": 'Build a ranked account-conquest plan from Leadbay data \u2014 the accounts worth\nactivating, each with a motif, a pitch and a checklist, ranked by the\nstrongest Leadbay signal. Every figure carries its source, and anything\nLeadbay can\'t measure is shown as OMITTED rather than estimated. Uses\n`leadbay_bulk_qualify_leads` and `leadbay_enrich_titles`. Trigger on\n"top 50 accounts to activate", "who should we go after".\n', "arguments": [{ "name": "count", "description": "Optional: how many accounts the plan should hold (default 50).", "required": false }, { "name": "territory", "description": "Optional: restrict the plan to a territory (e.g. 'Indre-et-Loire', 'R\xE9gion Ouest'). Sets geography on the Discover lens. A country is not a territory \u2014 this workspace already covers exactly one country.", "required": false }], "expected_calls": ["leadbay_account_status", "leadbay_get_qualification_questions", "leadbay_pull_leads", "leadbay_pull_followups", "leadbay_bulk_qualify_leads", "leadbay_qualify_status", "leadbay_scan_portfolio_signals", "leadbay_enrich_titles", "leadbay_bulk_enrich_status", "leadbay_account_history", "leadbay_artifact_kit", "leadbay_new_lens", "leadbay_adjust_audience"], "failure_modes": ["Invents, estimates or proxies a revenue-realized figure \u2014 the single worst failure. Leadbay does not hold what an account buys, and headcount, sector and lead score are NOT proxies for it.", "Sorts by cash-to-capture, synthesizing a revenue figure per account purely to make that ranking work. Leadbay has no revenue data: rank by the Leadbay signal, say so in the header, and never invent the key.", "Emits \u20AC figures with no provenance class, so modelled numbers read as measured fact in front of a paying client.", "Skips the PROVENANCE LEDGER, or drops un-sourceable fields from it instead of rendering them as OMITTED \u2014 which hides the gap.", "Fabricates registry/TAM counts (France or regional company counts) instead of querying the registry or marking the figure NOT COMPUTED. Leadbay does not proxy SIRENE.", "Invents the five qualification questions from this prompt's own recommendations instead of reading the org's actual questions via leadbay_get_qualification_questions.", "Leaves the deck's live layer dead \u2014 qualification and enrichment handles never wired in, so the pills and contacts stay empty while the deck still looks finished.", "Invents lead ids to make the qualification pills appear populated.", "Fabricates a plausible-sounding signal ('just won a public tender') for an account whose signals were never read. No signal read means an explicit dash.", "Assigns a motif outside the closed set of six, or assigns SAUVETAGE / PLAN DE COMPTE / MONT\xC9E EN GAMME / R\xC9VEIL from a lead score or sector when order history was never available.", "Labels Monitor membership as 'is a client' \u2014 Monitor is a Leadbay view whose membership is decided by lens scoring, not by whether the company ever bought anything.", "Launches paid enrichment on the whole plan without consent. Asking for a plan is not authorization to spend on 50 accounts.", "Re-launches enrichment from inside the built deck when a bulk handle already exists this session \u2014 double-spends the user's quota.", "Forces the interactive deck without offering it first, or ships the deck INSTEAD of a chat answer that stands on its own.", "Refuses the task because revenue data is missing, instead of delivering the conquest plan and naming what a cash-ranked version would need.", "Ends the turn without a ranked list of real accounts \u2014 gating the whole plan on a NON-blocking question (a MISSING territory, a missing lens, or a 3-vs-5 qualification-question gap) so the user gets a plan-of-a-plan. Only two things may stop delivery: an unresolvable company-identity mismatch, and a territory naming a foreign or supra-national scope. Every other open question rides alongside the delivered plan.", "Stops after the discovery contact preview to wait for enrichment consent, delivering no plan that turn \u2014 the ranked plan ships first; the paid reveal is offered alongside it.", "Renders a contact channel enrichment never returned (e.g. a phone link when only email was approved and revealed) instead of showing the returned channels and marking the rest omitted."] },
+  leadbay_top_accounts_to_activate: { "name": "leadbay_top_accounts_to_activate", "short_description": 'Build a ranked account-conquest plan from Leadbay data \u2014 the accounts worth\nactivating, each with a motif, a pitch and a checklist, ranked by the\nstrongest Leadbay signal. Every figure carries its source, and anything\nLeadbay can\'t measure is shown as OMITTED rather than estimated. Uses\n`leadbay_bulk_qualify_leads` and `leadbay_enrich_titles`. Trigger on\n"top 50 accounts to activate", "who should we go after".\n', "arguments": [{ "name": "count", "description": "Optional: how many accounts the plan should hold (default 50).", "required": false }, { "name": "territory", "description": "Optional: restrict the plan to a territory (e.g. 'Indre-et-Loire', 'R\xE9gion Ouest'). Sets geography on the Discover lens. A country is not a territory \u2014 this workspace already covers exactly one country.", "required": false }], "expected_calls": ["leadbay_account_status", "leadbay_get_qualification_questions", "leadbay_pull_leads", "leadbay_pull_followups", "leadbay_bulk_qualify_leads", "leadbay_qualify_status", "leadbay_scan_portfolio_signals", "leadbay_enrich_titles", "leadbay_bulk_enrich_status", "leadbay_account_history", "leadbay_artifact_kit", "leadbay_new_lens", "leadbay_adjust_audience"], "failure_modes": ["Invents, estimates or proxies a revenue-realized figure \u2014 the single worst failure. Leadbay does not hold what an account buys, and headcount, sector and lead score are NOT proxies for it.", "Sorts by cash-to-capture, synthesizing a revenue figure per account purely to make that ranking work. Leadbay has no revenue data: rank by the Leadbay signal, say so in the header, and never invent the key.", "Emits \u20AC figures with no provenance class, so modelled numbers read as measured fact in front of a paying client.", "Skips the PROVENANCE LEDGER, or drops un-sourceable fields from it instead of rendering them as OMITTED \u2014 which hides the gap.", "Fabricates registry/TAM counts (France or regional company counts) instead of querying the registry or marking the figure NOT COMPUTED. Leadbay does not proxy SIRENE.", "Invents the five qualification questions from this prompt's own recommendations instead of reading the org's actual questions via leadbay_get_qualification_questions.", "Leaves the deck's live layer dead \u2014 qualification and enrichment handles never wired in, so the pills and contacts stay empty while the deck still looks finished.", "Invents lead ids to make the qualification pills appear populated.", "Fabricates a plausible-sounding signal ('just won a public tender') for an account whose signals were never read. No signal read means an explicit dash.", "Assigns a motif outside the closed set of six, or assigns SAUVETAGE / PLAN DE COMPTE / MONT\xC9E EN GAMME / R\xC9VEIL from a lead score or sector when order history was never available.", "Labels Monitor membership as 'is a client' \u2014 Monitor is a Leadbay view whose membership is decided by lens scoring, not by whether the company ever bought anything.", "Launches enrichment on the whole plan without consent. Asking for a plan is not authorization to use quota on 50 accounts.", "Re-launches enrichment from inside the built deck when a bulk handle already exists this session \u2014 uses the user's quota twice.", "Forces the interactive deck without offering it first, or ships the deck INSTEAD of a chat answer that stands on its own.", "Refuses the task because revenue data is missing, instead of delivering the conquest plan and naming what a cash-ranked version would need.", "Ends the turn without a ranked list of real accounts \u2014 gating the whole plan on a NON-blocking question (a MISSING territory, a missing lens, or a 3-vs-5 qualification-question gap) so the user gets a plan-of-a-plan. Only two things may stop delivery: an unresolvable company-identity mismatch, and a territory naming a foreign or supra-national scope. Every other open question rides alongside the delivered plan.", "Stops after the discovery contact preview to wait for enrichment consent, delivering no plan that turn \u2014 the ranked plan ships first; the reveal is offered alongside it.", "Renders a contact channel enrichment never returned (e.g. a phone link when only email was approved and revealed) instead of showing the returned channels and marking the rest omitted."] },
   leadbay_work_campaign: { "name": "leadbay_work_campaign", "short_description": "Work a campaign as a real outreach session: pick the campaign,\nassess what the user has (phones / emails / coords), then PROPOSE\nthe right session mode (call sheet, email sheet, enrich titles\nfirst, map). After they pick, render \u2014 and as they dictate\noutcomes per lead, record both note + epilogue via\n`leadbay_report_outreach` in one round trip.\n", "arguments": [{ "name": "campaign", "description": "Campaign name (fuzzy match against your own campaigns) or campaign UUID. Omit to list and pick interactively.", "required": false }, { "name": "mode", "description": "Optional: skip the readiness-assessment proposal and jump directly into 'call_sheet' / 'email_sheet' / 'map' / 'enrich_first'. Omit (recommended) and let the prompt propose based on the data.", "required": false }], "expected_calls": ["leadbay_list_campaigns", "leadbay_campaign_call_sheet", "leadbay_enrich_titles", "leadbay_report_outreach"], "failure_modes": ["Renders the call sheet immediately without proposing the right mode \u2014 if 60% of leads have no contacts, calling is futile; enrich first. Always assess `readiness` first.", "Auto-renders the map widget without asking \u2014 maps are intrusive when the user just wants to scroll a list. Map mode is a proposed option, not a default.", "Proposes map mode after the user has previously said they don't like maps \u2014 check conversation memory before adding 'View on a map' to the options list.", "Calls `leadbay_campaign_progression` instead of `leadbay_campaign_call_sheet` \u2014 progression has counts but no phones / LinkedIn / call-ready data; the user can't actually dial from progression rows.", "Renders contacts WITHOUT making the phone number a `[bare](tel:URL)` link \u2014 on mobile that breaks one-tap calling, which is the whole point of the cheat sheet.", "Records outreach WITHOUT epilogue_status \u2014 leaves the lead's pipeline state unchanged; the rep then sees the same lead surfaced again next session.", "Records outreach WITHOUT verification \u2014 verification.source/ref is REQUIRED. For calls, pass `{source: 'user_confirmed', ref: <user's exact words>}`.", "Loops through ALL leads in a 50-lead campaign before recording any outreach \u2014 the call-then-record loop must be per-lead, not batched."] }
 };
 var PROMPT_CATALOG_HEADER = `This server exposes the following workflow prompts via \`prompts/list\` and \`prompts/get\`. Some MCP clients render them as slash commands; if your client does not, you (the agent) should invoke them directly via \`prompts/get\` when the user's request matches one of the triggers described below.`;
@@ -3627,7 +3629,7 @@ var LeadbayClient = class _LeadbayClient {
   // Cancelled while queued — nothing was ever put on the wire, which is what
   // makes this safe to report as "not sent" even for a write.
   cancelledBeforeSendError() {
-    return this.makeError("REQUEST_CANCELLED", "The request was cancelled before it was sent.", "Re-call the tool if you still want the result \u2014 nothing reached the API, so nothing was charged.");
+    return this.makeError("REQUEST_CANCELLED", "The request was cancelled before it was sent.", "Re-call the tool if you still want the result \u2014 nothing reached the API, so nothing ran.");
   }
   releaseSemaphore() {
     this.activeRequests--;
@@ -3967,6 +3969,11 @@ var LeadbayClient = class _LeadbayClient {
     if (status === 401) {
       return this.makeError("AUTH_EXPIRED", "Leadbay rejected this request (401)", retried ? "Tokens don't expire on a timer, so this isn't stale. Already auto-retried once and it 401'd again \u2014 usually a Leadbay-side hiccup, but can also mean the user logged out. Try again shortly, else report it." : "Tokens don't expire on a timer, so this isn't stale. This call wasn't auto-retried, so it's the first attempt \u2014 a Leadbay-side hiccup, or the user logged out. Try again once, else report it.", endpoint, null, status);
     }
+    const refusalCode = typeof parsed?.error === "string" ? parsed.error : parsed?.error?.code;
+    if (status === 429 && (refusalCode === "rate_limited" || refusalCode === "active_job_cap") && /^\/mcp\/(search|qualify)(\?|$)/.test(endpoint)) {
+      const wait = retryAfter ? `${retryAfter}s` : "a few minutes";
+      return this.makeError("QUOTA_EXCEEDED", refusalCode === "rate_limited" ? `Too many lead jobs started this hour \u2014 retry in ${wait}` : "Too many lead jobs running at once", (refusalCode === "rate_limited" ? `Leadbay starts at most 10 new lead jobs per org in a rolling hour. ` : `Leadbay runs at most 3 lead jobs per org at a time. `) + `This is a pacing limit, not credits: a top-up does not lift it, so do not offer one and do not talk about credits. Jobs already started keep running \u2014 read them with leadbay_lead_job_status meanwhile. ` + (refusalCode === "rate_limited" ? `Wait ${wait}, then retry the same call once. If it is refused again, the hour has not rolled over yet: wait longer rather than retrying in a loop.` : `Retry this call when one of the running jobs finishes.`), endpoint, retryAfter, status);
+    }
     if (status === 429 || status === 402 || parsed?.error === "quota_exceeded" || parsed?.error?.code === "quota_exceeded") {
       const hintBase = retryAfter ? `Wait ${retryAfter}s before retrying` : "Wait, then retry";
       return this.makeError(
@@ -4562,9 +4569,9 @@ Examples that should NOT invoke this tool (sound similar, route elsewhere):
 
 ## RENDER (quick)
 
-Report user + org, AND quota whenever readable \u2014 include quota even on a plain
-"what account?" ask. NEVER mention the lens unless asked (use
-\`last_requested_lens_name\`, never the id). SILENT on quota ONLY when
+Report user + org; show quota only when asked (quota / account status) or a
+window is exhausted. NEVER mention the lens unless asked (use
+\`last_requested_lens_name\`, never the id). SILENT on quota when
 \`quota_error\` set, \`unlimited_credits\` true, or quota null. Else render
 Daily/Weekly/Monthly from \`quota.user\` (fall back to \`quota.org\` if \`user\`
 absent) as \`$used / $cap (N% used) \xB7 resets\` (or a resource-count table when
@@ -4572,7 +4579,7 @@ absent) as \`$used / $cap (N% used) \xB7 resets\` (or a resource-count table whe
 
 ---
 
-Show the user's account state \u2014 admin rights, language, last-active lens, quota usage across daily/weekly/monthly windows, and whether the org's intelligence is mid-regeneration. **Show quota the way the web app does \u2014 a percentage-used + dollar-spend gauge per window, never raw "credits".** Each window in \`quota.<group>.spend[]\` carries \`current_units\` / \`max_units\` in dollar_cents (% used = the ratio, $ = \`/100\`); the \`quota.<group>.resources[]\` list gives the per-resource usage breakdown (\`count\`, plus \`max_units\` when a per-resource cap exists). **Pre-check the \`LENS_EXTRA_REFILL\` resource here before calling \`leadbay_extend_lens\`** \u2014 look in **\`quota.org.resources[]\`** first (admins), and fall back to **\`quota.user.resources[]\`** when \`quota.org\` is absent (non-admin callers only get the \`user\` group), matching the resource type **case-insensitively** (it may arrive as \`LENS_EXTRA_REFILL\` or \`lens_extra_refill\`). Its full requested batch must fit into the remaining daily quota or the call is rejected outright. Quota windows also hint at the user's consumption pace: heavy recent activity (ai_rescore / web_fetch near their window limits) is a signal that Leadbay will deliver a larger fresh batch next time the user logs back in, since batch size is paced by real consumption.
+Show the user's account state \u2014 admin rights, language, last-active lens, quota usage across daily/weekly/monthly windows, and whether the org's intelligence is mid-regeneration. **When you show quota, show it the way the web app does \u2014 a percentage-used + dollar-spend gauge per window, never raw "credits".** Each window in \`quota.<group>.spend[]\` carries \`current_units\` / \`max_units\` in dollar_cents (% used = the ratio, $ = \`/100\`); the \`quota.<group>.resources[]\` list gives the per-resource usage breakdown (\`count\`, plus \`max_units\` when a per-resource cap exists). **Pre-check the \`LENS_EXTRA_REFILL\` resource here before calling \`leadbay_extend_lens\`** \u2014 look in **\`quota.org.resources[]\`** first (admins), and fall back to **\`quota.user.resources[]\`** when \`quota.org\` is absent (non-admin callers only get the \`user\` group), matching the resource type **case-insensitively** (it may arrive as \`LENS_EXTRA_REFILL\` or \`lens_extra_refill\`). Its full requested batch must fit into the remaining daily quota or the call is rejected outright. Quota windows also hint at the user's consumption pace: heavy recent activity (ai_rescore / web_fetch near their window limits) is a signal that Leadbay will deliver a larger fresh batch next time the user logs back in, since batch size is paced by real consumption.
 
 **Top-ups always beat waiting.** When a quota window is hit, the user has two options: wait for the window reset (\`resets_at\` in each quota entry) OR top up AI credits. Top-ups clear the throttle IMMEDIATELY; they are not subject to the same window. When you tell the user about a 429 / quota exhaustion, ALWAYS surface both options \u2014 "wait until <reset>" or "top up now (I can generate the link)" \u2014 and let them pick. Never default-recommend "wait until tomorrow" when a 30-second top-up unblocks the same operation.
 
@@ -4610,7 +4617,7 @@ Some Leadbay tool responses include a \`_meta.notifications\` array listing **ba
 
 - **Previous turn / before an MCP restart, and the user has NOT asked about it** \u2014 don't poll for it in the background. Simply continue the conversation; the next time you call any tool, the completed-work entry appears in \`_meta.notifications\` (also on \`leadbay_account_status.notifications\`). This is the ambient push path \u2014 leave it to do its job. **But if the user explicitly asks for status or to "wait for it to finish"** (e.g. a multi-turn flow where a job was launched in a prior turn and this turn says "wait for enrichment to finish, then \u2026"), DO poll its status tool now until done, exactly as for a this-turn job below \u2014 the ambient push only surfaces *completed* work, so it can't answer a live "is it done / wait for it" request while the job is still running.
 - **This turn (you just launched it)** \u2014 the DEFAULT is: do NOT end your turn on the "launched" ack; stay active and poll the job's status tool in a loop until it reports done, then report the finished result yourself, rather than spinning forever or deferring the result to a later turn. (Two exceptions, detailed below: the user explicitly asked NOT to wait / to run it in the background; or it's a large qualification/import that's async by design \u2014 in those cases hand back the handle instead of looping.) Each status tool has its OWN terminal signal \u2014 poll until:
-  - \`leadbay_bulk_enrich_status\` \u2192 \`all_done:true\` \u2014 OR \`overall_progress.done\` holds steady across several SPACED polls (~15\u201330s apart) over at least ~90s\u20132 min of elapsed time (some contacts are unresolvable, so \`all_done\` can stay false forever). Don't call a plateau from the first few back-to-back reads \u2014 early on \`done\` sits flat while the backend spins up. Once the plateau is real, report what resolved and name what didn't.
+  - \`leadbay_bulk_enrich_status\` \u2192 \`all_done:true\` \u2014 OR \`overall_progress.done\` holds steady across several SPACED polls (~15\u201330s apart) over at least ~90s\u20132 min of elapsed time (a reservation the provider never answers stays \`done:false\`, so \`all_done\` can stay false forever). Don't call a plateau from the first few back-to-back reads \u2014 early on \`done\` sits flat while the backend spins up. Once the plateau is real, report what resolved and name what didn't.
   - \`leadbay_qualify_status\` \u2192 \`still_running\` is empty: every launched lead has finished or failed. (\`in_progress\` also reads \`false\` on the fast path, but it can be \`null\` on the legacy/fallback read \u2014 so treat an empty \`still_running\` as terminal on its own; only require \`in_progress:false\` when that field is actually present.) LIKE imports, large qualification runs are async by design: \`leadbay_bulk_qualify_leads\` defaults to \`wait_for_completion:false\` for \`count > 5\` or chained workflows because blocking can time out, and \`leadbay_qualify_status\` may take minutes/hours. So don't force a long polling loop on a big run \u2014 return the handle/progress and let completion arrive via \`_meta.notifications\` \u2014 UNLESS the user explicitly asked to wait, or it's a small run that finishes quickly. A small \`wait_for_completion:true\` run you can poll to \`still_running\` empty inline.
   - \`leadbay_import_status\` \u2192 \`status:"complete"\` (or \`"failed"\`). BUT imports are the exception to the stay-active loop: a large \`leadbay_import_leads({wait_for_completion:false})\` is meant to return a handle and resolve over minutes, and the tool does ONE refresh pass per call. Don't block the conversation looping on it \u2014 surface the returned progress/handle and let the completion arrive via \`_meta.notifications\` \u2014 UNLESS the user explicitly asked you to wait for the import, or it's a small import that finishes quickly.
 
@@ -4628,10 +4635,10 @@ Mirror the Leadbay web quota widget: three windows side by side \u2014 **Daily**
 $ cap** figure, with a per-resource usage breakdown underneath. **Never speak in
 raw "credits"** for quota \u2014 the unit is a percentage and a dollar spend.
 
-**Include the quota whenever it is readable** \u2014 as part of the default account
-answer, even when the user only asked "what account am I connected to?". The
-sole reason to omit it is the silence gate below (unreadable quota, or an
-unlimited account); it is NOT gated on the user explicitly asking for quota.
+**Show the quota only when it matters** \u2014 when the user asks about their quota,
+usage or account status, or when a window is exhausted and blocks what they
+asked for. A plain "what account am I connected to?" is answered with user +
+org alone. Even then, the silence gate below comes first.
 
 **Silence gate (check FIRST).** Render NOTHING about quota when any of these
 holds \u2014 do not mention quota at all, do not say "unreadable", never tell the user
@@ -4691,7 +4698,7 @@ days"), computed against now \u2014 mirroring the widget's "r\xE9initialis\xE9 d
 raw value is an ISO-8601 timestamp.
 
 **Top-up (optional, subordinate).** When \`quota.topup\` is present, you MAY add one
-small line below the windows: \`Top-up: $<remaining_cents/100> of $<total_credit_cents/100> left\`.
+small line below the windows: \`Top-up: $<(remaining_cents / 100).toFixed(2)> of $<(total_credit_cents / 100).toFixed(2)> left\`.
 Keep it secondary \u2014 the three window gauges are the headline. Omit when null.
 
 **Legend** (once, below): \`\` \`\u25B0\` used \xB7 \`\u25B1\` remaining \`\`.
@@ -4699,7 +4706,7 @@ Keep it secondary \u2014 the three window gauges are the headline. Omit when nul
 
 ---
 
-WHEN TO USE: at the start of a session to know what the agent can/can't do, after a 429 to explain to the user which resource window was exhausted and when it resets (and to offer the top-up alternative), and after the user signals a top-up so the agent can resume the interrupted workflow.
+WHEN TO USE: when the user asks about their account, quota or version, after a 429 to explain to the user which resource window was exhausted and when it resets (and to offer the top-up alternative), and after the user signals a top-up so the agent can resume the interrupted workflow.
 
 WHEN NOT TO USE: as a pre-flight gate before bulk ops \u2014 operations themselves return 429; this tool is for context, not gating. And: a recent quota snapshot showing "exhausted" is NOT a reason to refuse a write call when the user has just topped up \u2014 re-call this tool first, then proceed.
 `;
@@ -4923,7 +4930,7 @@ WHEN NOT TO USE: the user wants a plain data answer (route to leadbay_pull_leads
 `;
 var leadbay_bulk_enrich_status = `Check status + per-lead contacts for a bulk enrichment you previously launched via leadbay_enrich_titles. Pass the \`notification_id\` for the job counters in one call, and/or the \`lead_ids\` + \`titles\` + \`email\` / \`phone\` the launch returned for per-lead progress. \`lead_ids\` alone is a valid call and is the reliable one: the job lookup is a scan of your recent notifications, so an archived job may not be found, and an enrichment notification does not always carry counters (then the tool answers \`ENRICH_JOB_NO_COUNTERS\` with the backend's running/finished flag and asks for \`lead_ids\`) \u2014 but the leads always answer. When \`include_contacts=true\` (opt-in), includes each contact's email/phone_number/job_title/enrichment.done.
 
-WHEN TO USE: poll this REPEATEDLY after leadbay_enrich_titles returns a \`notification_id\`, staying active until the job is done \u2014 don't stop after one check, and don't hand the turn back to the user while progress is still climbing. "Done" = \`all_done:true\`, OR \`overall_progress.done\` has held steady across several SPACED polls (~15\u201330s apart) over at least ~90s\u20132 min of elapsed time (some contacts are unresolvable and never flip, so \`all_done\` can stay false forever \u2014 don't spin indefinitely). Do NOT declare a plateau from the first few back-to-back reads: right after launch, \`overall_progress.done\` can sit flat while the backend is still spinning the job up, so space your polls out and give it real elapsed time before treating a flat count as terminal. Also do NOT declare a plateau while the result carries \`partial_failures\` \u2014 a flat \`done\` there means a transient per-lead fetch error (e.g. a 429), NOT an unresolvable contact; keep polling (respecting any \`retry_after\`) or surface it as a temporary status failure, rather than reporting those leads as permanently unresolved. Default \`include_contacts=false\` for the cheap interim polls; set \`include_contacts=true\` on the read you report from to pull each lead's enriched contacts for the completion report.
+WHEN TO USE: poll this REPEATEDLY after leadbay_enrich_titles returns a \`notification_id\`, staying active until the job is done \u2014 don't stop after one check, and don't hand the turn back to the user while progress is still climbing. "Done" = \`all_done:true\`, OR \`overall_progress.done\` has held steady across several SPACED polls (~15\u201330s apart) over at least ~90s\u20132 min of elapsed time (a reservation the provider never answers stays \`done:false\`, so \`all_done\` can stay false forever \u2014 don't spin indefinitely; a contact with nothing found does flip, to \`done:true\` with \`credits_used:0\`). Do NOT declare a plateau from the first few back-to-back reads: right after launch, \`overall_progress.done\` can sit flat while the backend is still spinning the job up, so space your polls out and give it real elapsed time before treating a flat count as terminal. Also do NOT declare a plateau while the result carries \`partial_failures\` \u2014 a flat \`done\` there means a transient per-lead fetch error (e.g. a 429), NOT an unresolvable contact; keep polling (respecting any \`retry_after\`) or surface it as a temporary status failure, rather than reporting those leads as permanently unresolved. Default \`include_contacts=false\` for the cheap interim polls; set \`include_contacts=true\` on the read you report from to pull each lead's enriched contacts for the completion report.
 
 WHEN NOT TO USE: as a substitute for leadbay_research_lead_by_id \u2014 that already includes enriched contacts for a single lead.
 
@@ -4957,17 +4964,17 @@ a launcher only for a subset that never started, never for the whole batch:
   that are the ones that never started. Only then call
   \`leadbay_bulk_qualify_leads({leadIds, lensId})\` for exactly those ids. A lead
   that is merely slow looks identical to one that never launched over a few
-  polls, and re-launching it charges the user twice \u2014 when unsure, tell the user
+  polls, and re-launching it uses the user's quota twice \u2014 when unsure, tell the user
   rather than guess.
 
 
-## QUOTA \u2014 show where the user stands after the spend
+## QUOTA, NOT CREDITS
 
-Enrichment consumes QUOTA (the per-window allowance), not a separate credit wall. Once the job is done (all_done, or a plateau \u2014 see WHEN TO USE), show the user their refreshed quota: call \`leadbay_account_status\` and render the per-window quota it returns (the canonical surface). The result's \`credits_remaining\` field is **advisory internal context only \u2014 do NOT display it**: it comes from \`billing.ai_credits\` (a consumed counter, not remaining), so printing \`_(N credits remaining)_\` can show a fresh/quota-backed account a false "0 remaining." Never render a credits balance; the \`leadbay_account_status\` quota gauge is the only place the user's standing is shown. Do NOT report a "credits used" figure for this run either: the per-contact cost can't be scoped to this specific enrichment (a lead's contact list mixes in earlier runs), so any "X used" number would be misleading. Do the account_status refresh ONCE at completion \u2014 not on every in-progress poll.
+Enrichment consumes QUOTA (the per-window allowance), not a separate credit wall. The result's \`credits_remaining\` field is **advisory internal context only \u2014 do NOT display it**: it comes from \`billing.ai_credits\` (a consumed counter, not remaining), so printing \`_(N credits remaining)_\` can show a fresh/quota-backed account a false "0 remaining." Never render a credits balance; the \`leadbay_account_status\` quota gauge is the only place the user's standing is shown. Do NOT report a "credits used" figure for this run either: the per-contact cost can't be scoped to this specific enrichment (a lead's contact list mixes in earlier runs), so any "X used" number would be misleading.
 
 ## COMPLETION REPORT \u2014 what to tell the user when the job is done
 
-The result always carries \`overall_progress:{done,total,done_ratio}\` and, with \`include_contacts:true\`, \`leads[]\` each with contacts' \`email\` / \`phone_number\` / \`job_title\` / \`enrichment.done\`. \`bulk_progress:{total_count,success_count,failure_count,quota_hit_count}\` is present only when you passed a \`notification_id\` AND the job was found; derive counts from \`overall_progress\` rather than assuming \`bulk_progress\` is there. With \`lead_ids\`, each entry carries \`enrichment_progress:{done,total}\` \u2014 \`done\` counts only contacts whose REQUESTED channel has landed, scoped to the \`titles\` this run enriched, so a lead's pre-existing CFO email cannot inflate a CEO run. A contact counts as done only when the REQUESTED channel actually landed \u2014 for a phone run, \`enrichment.done:true\` with no \`phone_number\` is NOT done (the contact may have been email-enriched earlier); read \`email\` / \`phone_number\` against the requested channels, don't rely on the \`enrichment.done\` flag alone (\`overall_progress\` already accounts for this). \`include_contacts\` returns each lead's FULL contact list (it fans out through \`leadbay_get_contacts\`), so it can include contacts of other roles that were enriched in earlier runs \u2014 filter your report to the \`titles\` this bulk enriched (match each contact's \`job_title\`), don't attribute a pre-existing email of an unrelated role to this run. Report it yourself in the SAME turn, without a reprompt and without deferring to a scheduled re-check: name which of the just-enriched contacts now have emails / phones, the done/total counts, and \u2014 if \`bulk_progress\` is present \u2014 any \`quota_hit_count\` (if non-zero, say some contacts were skipped because the quota window was exhausted, and point to \`leadbay_account_status\` for the wait-or-top-up choice). If you stopped on a plateau (not \`all_done\`), say so plainly \u2014 report the resolved contacts and name the ones that didn't resolve, keyed to the requested channel and the returned fields (no \`email\` \u2192 "no email found"; no \`phone_number\` \u2192 "no phone number found") \u2014 rather than implying the job fully finished. Then show refreshed quota via \`leadbay_account_status\` (see QUOTA above); do NOT print a credits-remaining line.
+The result always carries \`overall_progress:{done,total,done_ratio}\` and, with \`include_contacts:true\`, \`leads[]\` each with contacts' \`email\` / \`phone_number\` / \`job_title\` / \`enrichment.done\`. \`bulk_progress:{total_count,success_count,failure_count,quota_hit_count}\` is present only when you passed a \`notification_id\` AND the job was found; derive counts from \`overall_progress\` rather than assuming \`bulk_progress\` is there. With \`lead_ids\`, each entry carries \`enrichment_progress:{done,total}\` \u2014 \`total\` counts the reservations this run made (scoped to the \`titles\` it enriched, and to reservations whose \`enrichment.email_requested\` / \`phone_requested\` match the \`email\` / \`phone\` you pass, so a lead's pre-existing CFO email cannot inflate a CEO run) and \`done\` counts those that have settled (\`enrichment.done:true\`, found or not). The \`enrichment\` record sits on the \`source:"paid"\` entry and never carries \`email\` / \`phone_number\`: the revealed values land on the same person's \`source:"org"\` entry (match on name and \`job_title\`). Read what landed from the org entries \u2014 an org entry with \`email\` = email found, with \`phone_number\` = phone found \u2014 and a settled paid entry with \`credits_used:0\` = nothing found for that person. \`include_contacts\` returns each lead's FULL contact list (it fans out through \`leadbay_get_contacts\`), so it can include contacts of other roles that were enriched in earlier runs \u2014 filter your report to the \`titles\` this bulk enriched (match each contact's \`job_title\`), don't attribute a pre-existing email of an unrelated role to this run. Report it yourself in the SAME turn, without a reprompt and without deferring to a scheduled re-check: name which of the just-enriched contacts now have emails / phones, the done/total counts, and \u2014 if \`bulk_progress\` is present \u2014 any \`quota_hit_count\` (if non-zero, say some contacts were skipped because the quota window was exhausted, and point to \`leadbay_account_status\` for the wait-or-top-up choice). If you stopped on a plateau (not \`all_done\`), say so plainly \u2014 report the resolved contacts and name the ones that didn't resolve, keyed to the requested channel and the returned fields (no \`email\` \u2192 "no email found"; no \`phone_number\` \u2192 "no phone number found") \u2014 rather than implying the job fully finished. Do NOT print a credits-remaining line.
 `;
 var leadbay_bulk_qualify_leads = `Pick the next N unqualified leads in the active lens and qualify them (run AI rescore + web fetch). Pass \`wait_for_completion:false\` to return quickly with \`{status:'running', notification_id}\`; poll leadbay_qualify_status with that id. With \`wait_for_completion\` omitted/true, the legacy behavior polls until the answers are populated or a budget is exhausted. Already-qualified leads (those with a non-null \`ai_agent_lead_score\`) are silently no-ops on the backend, so this composite paginates past them to find fresh candidates. On 429 mid-fanout, stops launching but keeps polling already-launched leads.
 
@@ -4984,7 +4991,7 @@ WHEN NOT TO USE: to qualify a single specific lead \u2014 that's leadbay_qualify
 Leadbay has no cancel. Once \`leadbay_enrich_titles\`, \`leadbay_bulk_qualify_leads\`,
 \`leadbay_import_leads\` or \`leadbay_import_and_qualify\` has returned a launched or
 running result, that work is queued on Leadbay and runs to completion, and the
-quota it costs is already committed. A discovery, preview or \`dry_run\` result
+quota it uses is already committed. A discovery, preview or \`dry_run\` result
 launched nothing and is not covered here.
 
 The user cancelling in the chat, a request timeout, or a closed stream stops YOUR
@@ -4992,7 +4999,7 @@ waiting, never the job. \`cancelled: true\` means we stopped watching, not that
 work stopped. What to do next depends on what you are holding:
 
 - **A handle.** Poll the status tool with it, and do not launch the work that
-  handle covers a second time \u2014 that spends the quota again on the same rows.
+  handle covers a second time \u2014 that uses the quota again on the same rows.
   \`leadbay_import_status\` takes \`importIds\`, so pass the values of \`import_ids\`
   under that name. A qualification started by \`leadbay_import_and_qualify\` has no
   notification of its own: resume it with
@@ -5006,7 +5013,7 @@ work stopped. What to do next depends on what you are holding:
   finished. Calling the same tool again with the same arguments will usually hand
   back the job already launched rather than starting a second one, but that guard
   is in-memory, five minutes, and per process, so it is best-effort \u2014 say what you
-  are about to re-run before you spend the user's quota on it.
+  are about to re-run before you use the user's quota on it.
 
 
 This tool MUTATES state. The caller (agent or human-in-the-loop) is responsible for confirming intent before invocation; the MCP server does not soft-prompt for confirmation. See \`annotations.destructiveHint\`.
@@ -5496,29 +5503,29 @@ WHEN NOT TO USE: for bulk enrichment by job title across many leads \u2014 use l
 ## A launched job cannot be stopped, and this tool has no retry guard
 
 Leadbay has no cancel. Once this call returns having actually launched, the work
-is queued on Leadbay and runs to completion, and the quota it costs is already
-committed. A \`dry_run\` result reached no backend and spent nothing. The user
+is queued on Leadbay and runs to completion, and the quota it uses is already
+committed. A \`dry_run\` result reached no backend and used nothing. The user
 cancelling in the chat, a request timeout, or a closed stream stops YOUR waiting,
 never the job.
 
 Unlike the composite launchers, this tool has **no double-launch guard**: calling
-it again always issues a new paid launch, even seconds later with identical
+it again always issues a new launch that uses quota again, even seconds later with identical
 arguments. So when a call returns nothing at all, do not simply retry. Read the
 record back first \u2014 \`leadbay_research_lead_by_id\` or \`leadbay_get_contacts\` for a
 lead, \`leadbay_account_status\` for background work that has since finished \u2014 to
 see whether the launch already landed, and tell the user what you are about to
-spend before spending it again.
+run before running it again.
 
 
 ## QUOTA, NOT CREDITS
 
-Enrichment is gated by QUOTA (the per-window allowance in \`leadbay_account_status\`), not a credit balance. **Never pre-refuse because a credit number looks low or zero** \u2014 a freemium/fresh account with quota left can enrich even when its credit counter reads 0. The reveal either fits the remaining quota or the backend returns 429 (\`quota_exceeded\`); only THEN surface the exhausted window + wait-or-top-up choice. The \`credits_remaining\` field on the result is **advisory internal context only \u2014 do NOT display it**. Because it can read \`0\` on an account that still has quota, printing \`_(N credits remaining)_\` would falsely tell the user they're out. Do not render a credits balance at all; if the user asks where they stand, call \`leadbay_account_status\` and show the quota gauge instead. The actual per-contact cost (\`enrichment.credits_used\`) appears on the contact after enrichment.
+Enrichment is gated by QUOTA (the per-window allowance in \`leadbay_account_status\`), not a credit balance. **Never pre-refuse because a credit number looks low or zero** \u2014 a freemium/fresh account with quota left can enrich even when its credit counter reads 0. The reveal either fits the remaining quota or the backend returns 429 (\`quota_exceeded\`); only THEN surface the exhausted window + wait-or-top-up choice. The \`credits_remaining\` field on the result is **advisory internal context only \u2014 do NOT display it**. Because it can read \`0\` on an account that still has quota, printing \`_(N credits remaining)_\` would falsely tell the user they're out. Do not render a credits balance at all; if the user asks where they stand, call \`leadbay_account_status\` and show the quota gauge instead.
 
 **Channels: when the user asks to enrich a contact without naming a channel, confirm scope via \`ask_user_input_v0\`** \u2014 \`"Enrich email only, or email + phone? (phone uses more quota)"\` \u2192 \`["Email only", "Email + phone"]\` \u2014 then pass the chosen \`email\`/\`phone\` flags. Skip the question only if they already said which channel(s) they want.
 
 This tool MUTATES state. The caller (agent or human-in-the-loop) is responsible for confirming intent before invocation; the MCP server does not soft-prompt for confirmation. See \`annotations.destructiveHint\`.
 `;
-var leadbay_enrich_titles = `Order contact enrichments by job title across many leads. Contacts are NOT returned by default with a lead (Leadbay keeps enrichment out-of-band to control cost); the agent requests them on demand via this tool when it's ready to actually reach out. Two modes: (A) NO \`titles\` param \u2014 returns the available titles + Leadbay's \`title_suggestions\` + \`auto_included_titles\` + a count of enrichable contacts, so the agent can ask the user which titles to enrich. (B) \`titles\` given \u2014 calls preview, then launches if there's anything enrichable. On 429 returns \`{status:'quota_exceeded'}\` cleanly. Selection lifecycle is wrapped in a try/finally so the user's selection is left clean even on error.
+var leadbay_enrich_titles = `Order contact enrichments by job title across many leads. Contacts are NOT returned by default with a lead (Leadbay keeps enrichment out-of-band); the agent requests them on demand via this tool when it's ready to actually reach out. Two modes: (A) NO \`titles\` param \u2014 returns the available titles + Leadbay's \`title_suggestions\` + \`auto_included_titles\` + a count of enrichable contacts, so the agent can ask the user which titles to enrich. (B) \`titles\` given \u2014 calls preview, then launches if there's anything enrichable. On 429 returns \`{status:'quota_exceeded'}\` cleanly. Selection lifecycle is wrapped in a try/finally so the user's selection is left clean even on error.
 
 WHEN TO USE: as the agent's go-to enrichment entry point, immediately before proposing outreach.
 
@@ -5529,7 +5536,7 @@ WHEN NOT TO USE: to enrich a single named contact \u2014 that's leadbay_enrich_c
 Leadbay has no cancel. Once \`leadbay_enrich_titles\`, \`leadbay_bulk_qualify_leads\`,
 \`leadbay_import_leads\` or \`leadbay_import_and_qualify\` has returned a launched or
 running result, that work is queued on Leadbay and runs to completion, and the
-quota it costs is already committed. A discovery, preview or \`dry_run\` result
+quota it uses is already committed. A discovery, preview or \`dry_run\` result
 launched nothing and is not covered here.
 
 The user cancelling in the chat, a request timeout, or a closed stream stops YOUR
@@ -5537,7 +5544,7 @@ waiting, never the job. \`cancelled: true\` means we stopped watching, not that
 work stopped. What to do next depends on what you are holding:
 
 - **A handle.** Poll the status tool with it, and do not launch the work that
-  handle covers a second time \u2014 that spends the quota again on the same rows.
+  handle covers a second time \u2014 that uses the quota again on the same rows.
   \`leadbay_import_status\` takes \`importIds\`, so pass the values of \`import_ids\`
   under that name. A qualification started by \`leadbay_import_and_qualify\` has no
   notification of its own: resume it with
@@ -5551,7 +5558,7 @@ work stopped. What to do next depends on what you are holding:
   finished. Calling the same tool again with the same arguments will usually hand
   back the job already launched rather than starting a second one, but that guard
   is in-memory, five minutes, and per process, so it is best-effort \u2014 say what you
-  are about to re-run before you spend the user's quota on it.
+  are about to re-run before you use the user's quota on it.
 
 
 ## ENRICHMENT CONSUMES QUOTA \u2014 the model to reason with
@@ -5560,25 +5567,25 @@ Each email reveal and each phone reveal **consumes quota** (the per-window daily
 
 ## CONSENT \u2014 email is the default channel; phone is opt-in; never launch silently
 
-The \`email\` channel defaults **ON**; \`phone\` defaults **OFF**. A bare "enrich these titles" is **NOT** consent to spend quota.
+The \`email\` channel defaults **ON**; \`phone\` defaults **OFF**. A bare "enrich these titles" is **NOT** consent to use quota.
 
 **When the user asks to enrich without naming channels, ASK which channels via \`ask_user_input_v0\`** \u2014 email is included by default, so the real question is whether to add phone: \`"Enrich email only, or email + phone? (phone reveals use more quota)"\` \u2192 \`["Email only", "Email + phone"]\`. Then launch with the chosen channels (\`email:true\` always; \`phone:true\` if they picked email + phone). Only skip this question if the user already named the channel(s) explicitly ("just emails", "get their phone numbers too", etc.).
 
-**To preview with ZERO spend risk on ANY host, pass \`dry_run:true\`** (or \`confirm:false\`). Either returns \`enrichable_contacts\` + the \`would_launch\` channels and launches nothing \u2014 guaranteed, regardless of host. Use this as your first call. Then surface the volume, get the user's explicit go-ahead (and channel choice, above), and re-call with \`confirm:true\` (and explicit \`email:true\`/\`phone:true\` for the channels they chose) to launch.
+**To preview with nothing launched on ANY host, pass \`dry_run:true\`** (or \`confirm:false\`). Either returns \`enrichable_contacts\` + the \`would_launch\` channels and launches nothing \u2014 guaranteed, regardless of host. Use this as your first call. Then surface the volume, get the user's explicit go-ahead (and channel choice, above), and re-call with \`confirm:true\` (and explicit \`email:true\`/\`phone:true\` for the channels they chose) to launch.
 
-Do NOT rely on a bare call (no \`confirm\`, no \`dry_run\`, no channels) as a "safe preview": on an elicitation-capable host it asks the user and withholds on decline (\`mode:"needs_confirmation"\`), but on a host WITHOUT elicitation (some direct/embedded callers) a bare call launches the default email spend directly. If you're unsure whether the host can elicit, use \`dry_run:true\`/\`confirm:false\` for the preview. Passing \`email:true\`/\`phone:true\` (or \`confirm:true\`) always counts as consent and launches.
+Do NOT rely on a bare call (no \`confirm\`, no \`dry_run\`, no channels) as a "safe preview": on an elicitation-capable host it asks the user and withholds on decline (\`mode:"needs_confirmation"\`), but on a host WITHOUT elicitation (some direct/embedded callers) a bare call launches the default email reveal directly. If you're unsure whether the host can elicit, use \`dry_run:true\`/\`confirm:false\` for the preview. Passing \`email:true\`/\`phone:true\` (or \`confirm:true\`) always counts as consent and launches.
 
-## SHOW WHAT WILL RUN, AND WHERE QUOTA STANDS
+## SHOW WHAT WILL RUN
 
-**BEFORE (confirm before launching).** The discover / preview_only / dry_run modes return \`enrichable_contacts\` (the volume that would be enriched). Tell the user plainly: **"This will enrich {enrichable_contacts} contacts (email + phone reveals consume quota)."** then confirm the channels + go-ahead via \`ask_user_input_v0\` before launching. Do NOT quote an exact cost or a "credits" figure \u2014 the per-reveal rate is backend-side and enrichment is gated by quota, not a credit balance. The \`credits_remaining\` field is advisory context only; never present it as a spend gate and never refuse based on it.
+**BEFORE (confirm before launching).** The discover / preview_only / dry_run modes return \`enrichable_contacts\` (the volume that would be enriched). Tell the user plainly: **"This will enrich {enrichable_contacts} contacts (email + phone reveals consume quota)."** then confirm the channels + go-ahead via \`ask_user_input_v0\` before launching. Do NOT quote an exact cost or a "credits" figure \u2014 the per-reveal rate is backend-side and enrichment is gated by quota, not a credit balance. The \`credits_remaining\` field is advisory context only; never present it as a gate and never refuse based on it.
 
-**AFTER (show refreshed quota).** Once the job is done (see the STAY ACTIVE section below), call \`leadbay_account_status\` and show the refreshed per-window quota \u2014 the canonical surface \u2014 so the user sees the usage they just consumed. Do NOT invent a "credits used" figure for the run (per-run cost can't be scoped reliably \u2014 a lead's contacts mix earlier enrichments).
+**AFTER.** Do NOT invent a "credits used" figure for the run (per-run cost can't be scoped reliably \u2014 a lead's contacts mix earlier enrichments).
 
 ## AFTER LAUNCH \u2014 STAY ACTIVE UNTIL DONE
 
-When a launch returns \`mode:"launched"\` with a \`notification_id\`, the enrichment runs ASYNC on the backend \u2014 the tool returns immediately, before any email/phone is attached. **Unless the user explicitly said to start it in the background / not to wait** (e.g. "kick it off, I'll check later", "don't wait for it"), stay active and report in-turn \u2014 do NOT end your turn on the ack, and do NOT say "I'll let you know when it's done." (If the user DID ask you not to wait, honor that: hand back the \`notification_id\` and a one-line "running \u2014 you can ask any time". Only promise that completion will auto-surface via \`_meta.notifications\` when the launch returned a non-null \`notification_id\`; if \`notification_id\` is null (the nullable-backend path), say instead that you'll re-check when asked / they should ask again later \u2014 nothing surfaces automatically without a notification id. Don't force a poll loop against explicit intent.) In the default (stay-active) case: call \`leadbay_bulk_enrich_status({notification_id})\` in a loop, re-polling until the job is done (small batches typically finish in under ~2 min). Pass \`include_contacts:true\` on the read you intend to report from, so you get each lead's enriched contacts back. Note that \`include_contacts\` returns each lead's FULL contact list (it fans out through \`leadbay_get_contacts\`), which can include contacts of OTHER roles that were already enriched in earlier runs \u2014 so **filter your report to the \`titles\` you just enriched** (match each contact's \`job_title\` to the requested titles). Don't present a pre-existing CFO/Sales email as part of this CEO/Owner/Manager run. Then \u2014 on your own, without waiting for the user to reprompt \u2014 report the enrichment: which of the just-enriched contacts now have emails / phones, and the counts from \`overall_progress\` (\`done\`/\`total\`). \`leadbay_bulk_enrich_status\` also returns \`bulk_progress.success_count\` / \`failure_count\` / \`quota_hit_count\` on the notification fast path \u2014 use those when present, but the per-lead path returns \`overall_progress\` only, so don't assume \`bulk_progress\` exists (see the status tool's COMPLETION REPORT). Then show refreshed quota via \`leadbay_account_status\` (see AFTER above).
+When a launch returns \`mode:"launched"\` with a \`notification_id\`, the enrichment runs ASYNC on the backend \u2014 the tool returns immediately, before any email/phone is attached. **Unless the user explicitly said to start it in the background / not to wait** (e.g. "kick it off, I'll check later", "don't wait for it"), stay active and report in-turn \u2014 do NOT end your turn on the ack, and do NOT say "I'll let you know when it's done." (If the user DID ask you not to wait, honor that: hand back the \`notification_id\` and a one-line "running \u2014 you can ask any time". Only promise that completion will auto-surface via \`_meta.notifications\` when the launch returned a non-null \`notification_id\`; if \`notification_id\` is null (the nullable-backend path), say instead that you'll re-check when asked / they should ask again later \u2014 nothing surfaces automatically without a notification id. Don't force a poll loop against explicit intent.) In the default (stay-active) case: call \`leadbay_bulk_enrich_status({notification_id})\` in a loop, re-polling until the job is done (small batches typically finish in under ~2 min). Pass \`include_contacts:true\` on the read you intend to report from, so you get each lead's enriched contacts back. Note that \`include_contacts\` returns each lead's FULL contact list (it fans out through \`leadbay_get_contacts\`), which can include contacts of OTHER roles that were already enriched in earlier runs \u2014 so **filter your report to the \`titles\` you just enriched** (match each contact's \`job_title\` to the requested titles). Don't present a pre-existing CFO/Sales email as part of this CEO/Owner/Manager run. Then \u2014 on your own, without waiting for the user to reprompt \u2014 report the enrichment: which of the just-enriched contacts now have emails / phones, and the counts from \`overall_progress\` (\`done\`/\`total\`). \`leadbay_bulk_enrich_status\` also returns \`bulk_progress.success_count\` / \`failure_count\` / \`quota_hit_count\` on the notification fast path \u2014 use those when present, but the per-lead path returns \`overall_progress\` only, so don't assume \`bulk_progress\` exists (see the status tool's COMPLETION REPORT).
 
-**"Done" = \`all_done:true\` OR the resolvable work has plateaued.** Keep polling while \`overall_progress.done\` is still climbing. But \`total\` counts every matching contact, and some (unresolvable titles, contacts with no findable email) never flip to done \u2014 so a job can sit below 100% with \`all_done:false\` forever. A plateau is only real once the job has had time to run: do NOT declare it from the first few back-to-back reads (early on \`done\` can sit at its initial value while the backend is still spinning the job up). Give it at least ~90s\u20132 min of actual elapsed polling \u2014 space your polls out (~15\u201330s apart) rather than firing them back-to-back \u2014 and only treat the set as complete when \`overall_progress.done\` has held steady across several spaced polls over that window. Then stop polling and report what resolved, naming the ones that didn't. Key the "didn't resolve" wording off the channels the user actually requested and the returned contact fields (contacts carry \`email\` and \`phone_number\`) \u2014 a contact enriched for phone that came back with no \`phone_number\` is "no phone number found", one with no \`email\` is "no email found", email+phone that got neither is "no contact details found"; if \`quota_hit_count\` is non-zero say those were skipped because the quota window was exhausted. Do NOT hard-label every non-success as "no email found" when phone was requested. Do NOT spin indefinitely waiting for \`all_done\` on contacts the engine won't resolve, and do NOT \`ScheduleWakeup\` / defer the finished list to a later turn \u2014 deliver the resolved results in THIS reply.
+**"Done" = \`all_done:true\` OR the resolvable work has plateaued.** Keep polling while \`overall_progress.done\` is still climbing. But a reservation the provider never answers stays \`done:false\` \u2014 so a job can sit below 100% with \`all_done:false\` forever (a contact with no findable email does flip, to \`done:true\` with \`credits_used:0\`). A plateau is only real once the job has had time to run: do NOT declare it from the first few back-to-back reads (early on \`done\` can sit at its initial value while the backend is still spinning the job up). Give it at least ~90s\u20132 min of actual elapsed polling \u2014 space your polls out (~15\u201330s apart) rather than firing them back-to-back \u2014 and only treat the set as complete when \`overall_progress.done\` has held steady across several spaced polls over that window. Then stop polling and report what resolved, naming the ones that didn't. Key the "didn't resolve" wording off the channels the user actually requested and the returned contact fields (contacts carry \`email\` and \`phone_number\`) \u2014 a contact enriched for phone that came back with no \`phone_number\` is "no phone number found", one with no \`email\` is "no email found", email+phone that got neither is "no contact details found"; if \`quota_hit_count\` is non-zero say those were skipped because the quota window was exhausted. Do NOT hard-label every non-success as "no email found" when phone was requested. Do NOT spin indefinitely waiting for \`all_done\` on contacts the engine won't resolve, and do NOT \`ScheduleWakeup\` / defer the finished list to a later turn \u2014 deliver the resolved results in THIS reply.
 
 The \`_meta.notifications\` push is the FALLBACK for a job launched in an EARLIER turn or across an MCP restart \u2014 it is NOT a reason to stop early on a job you just launched this turn.
 
@@ -5720,8 +5727,8 @@ Examples that should NOT invoke this tool (sound similar, route elsewhere):
 
 3-col table of delivered leads in returned order: col 1 = 10-segment fit
 bar + linked company \xB7 location \xB7 size; col 2 = why-fits \u226420 words; col 3
-= contact + purchased channels. ALWAYS close with the honest funnel line
-(matched/examined/delivered/stop reason/spend) \u2014 especially on 0
+= contact + found channels. ALWAYS close with the honest funnel line
+(matched/examined/delivered/stop reason) \u2014 especially on 0
 delivered. Full algorithm below.
 
 ---
@@ -5729,19 +5736,19 @@ delivered. Full algorithm below.
 Submit a net-new lead search: the backend matches an ICP seed against the full
 company universe, applies hard filters, skips what the org already knows
 (\`novelty: org\`), optionally qualifies against the org's own intelligence
-(questions, tags, ideal buyer profile \u2014 frozen at submit), and optionally buys
+(questions, tags, ideal buyer profile \u2014 frozen at submit), and optionally reveals
 contact channels. Polls up to \`wait_seconds\` (default 45); a longer job returns
 \`still_running\` + \`next_poll\` \u2014 hand off to \`leadbay_lead_job_status\`. Jobs run
 \u226430 min, results kept 30 days.
 
-**Free vs paid \u2014 never spend silently.** Default (\`qualify: false\`,
+**Free vs usage quota \u2014 never use quota silently.** Default (\`qualify: false\`,
 \`channels: []\`) is FREE: company profile + fit score + cached research +
-contact identity. Paid: \`qualify: true\` (~94 cost_cents per candidate
-EXAMINED, capped by \`exploration_cap\`/\`max_cost\`) and \`channels\` (email 25c /
-phone 250c, success-only). Enforced in code: a paid call is WITHHELD unless it
+contact identity. \`qualify: true\` (per candidate EXAMINED, capped by
+\`exploration_cap\`/\`max_cost\`) and \`channels\` (only when a value is found) draw
+on the org's usage quota; nothing is invoiced. Enforced in code: such a call is WITHHELD unless it
 carries \`confirm: true\` \u2014 nothing is submitted and you get
 \`mode: "needs_confirmation"\` with a real quote to show the user. Re-call with
-\`confirm: true\` on their go-ahead ("spend / get their emails" counts).
+\`confirm: true\` on their go-ahead ("go ahead / get their emails" counts).
 \`confirm: false\` vetoes. Free needs no consent. **Preview free first** \u2014
 reshaping an off-profile seed is free, exploring it with \`qualify: true\` is
 not.
@@ -5749,7 +5756,7 @@ not.
 **Ad-hoc exclusions ("no chains") are enforced by NO tier** \u2014 \`filters\` has no
 exclusion key, and \`qualify\` scores against the org's FROZEN questions and IBP,
 which need not mention chains; the seed's inverse only shifts ranking.
-Violators can survive, be paid for and be delivered \u2014 post-filter them yourself
+Violators can survive, use quota and be delivered \u2014 post-filter them yourself
 and say the tier didn't enforce it. Durable enforcement \u2192
 \`leadbay_refine_prompt\`.
 
@@ -5796,9 +5803,9 @@ measured:
 
 **Parameter notes**
 - \`request_id\` (REQUIRED) is the retry contract: SAME value retrying the same
-  ask (same live job, no double spend); NEW for a changed ask. Derive from ask
+  ask (same live job, no double launch); NEW for a changed ask. Derive from ask
   + archetype + date: \`gyms-dallas-2026-07-28\`.
-- Never lower \`min_ai_score\` together with \`channels\` \u2014 that buys emails for
+- Never lower \`min_ai_score\` together with \`channels\` \u2014 that reveals emails for
   leads the AI just scored as junk.
 - \`count\` \u2264 50; \u22643 active jobs/org; \u226410 submits/hour (429 + Retry-After \u2014
   wait, don't hammer).
@@ -5838,7 +5845,7 @@ when nothing was delivered.
 **Column 3 \u2014 Contact**
 
 - \`[Name](linkedin) \xB7 role\` (linked name mandatory when a LinkedIn URL
-  exists; plain name otherwise). Below it, the PURCHASED channels only:
+  exists; plain name otherwise). Below it, the FOUND channels only:
   \`\u2709 value\` / \`\u260E value\` inline as plain text (they auto-linkify).
 - Channel statuses: \`delivered\` \u2192 show value; \`already_owned\` \u2192 value +
   *(already yours)*; \`masked\` \u2192 "on file \u2014 reveal via channels";
@@ -5849,16 +5856,17 @@ when nothing was delivered.
 
 **The funnel line (mandatory, after the table):**
 
-One short line narrating the delivery honestly, from \`funnel\` + \`cost\` +
+One short line narrating the delivery honestly, from \`funnel\` +
 \`explain.scope_notes\`:
 
 > Matched N \xB7 examined E \xB7 qualified Q \xB7 disqualified D \u2192 **delivered X of
-> the Y asked** \xB7 stopped: <stop_reason in plain words> \xB7 spent C.CC.
+> the Y asked** \xB7 stopped: <stop_reason in plain words>.
 
-**Money: divide, then symbol.** Every amount (\`cost.spent\`,
-\`estimated_cost.max\`, quotes) is \`cost_cents\` \u2014 divide by 100, two decimals,
-so \`165\` renders \`1.65\`, NEVER \`165.00\`. Symbol from the account region: US
-\`$\`, France \`\u20AC\`, unknown \u2192 bare. Never hard-code \`$\`: it misstates a charge.
+**No money, anywhere.** \`cost.*\`, \`estimated_cost.max\` and quotes are internal
+usage units. Never render them, never convert them to a currency, never call
+them a charge: the user's plan or top-up covers this work, and a price reads as
+a bill. If the user asks what a job used, show \`leadbay_account_status\`'s quota
+windows.
 
 "of the Y asked" needs \`summary.items_requested\`, which submits carry but a
 later \`leadbay_lead_job_status\` snapshot does not. Without it write **delivered
@@ -5866,7 +5874,7 @@ X** and stop \u2014 never back-fill Y from \`matched\`/\`examined\` (they count
 candidates), never guess it.
 
 Plain-word stop reasons: \`target_reached\` \u2192 omit (success), \`pool_exhausted\` \u2192
-"ran out of matching candidates", \`max_cost\` \u2192 "hit the cost cap", \`quota\` \u2192
+"ran out of matching candidates", \`max_cost\` \u2192 "hit the job's usage cap", \`quota\` \u2192
 "hit an org quota", \`time_budget\` \u2192 "hit the 30-min time budget".
 
 **When \`delivered\` is 0**: NEVER say just "no results". Render no table; give
@@ -5891,7 +5899,7 @@ table \`Ref \u2192 Outcome\` translating \`status_reason\` to plain words:
 \`low_confidence_identity\` \u2192 "couldn't safely match \u2014 check \`resolution.alternatives\`",
 \`no_matching_contact\` \u2192 "no contact with the requested title",
 \`disqualified\` \u2192 "evaluated: does not fit" (evidence is in the item when owned),
-\`enrichment_failed\` \u2192 "channel could not be sourced (not billed)".
+\`enrichment_failed\` \u2192 "channel could not be sourced (no quota used)".
 
 **\`items_truncated\`**: rows are a PREFIX, not the batch. Say so, and offer
 \`leadbay_lead_job_status(job_id, since: next_since)\` for the rest.
@@ -5940,10 +5948,10 @@ Pick the 2-3 options that match what actually happened \u2014 never all seven:
 | Observation | Suggest | Calls |
 |---|---|---|
 | Job still running (\`still_running: true\`) | "Check on it in ~1 min" | leadbay_lead_job_status(job_id, wait_seconds: 60) |
-| Free run delivered on-profile leads | "Qualify these N against your criteria (paid \u2014 \`dry_run\` first)" | leadbay_qualify_leads(prior_deliveries: {job_id}) |
+| Free run delivered on-profile leads | "Qualify these N against your criteria (uses quota \u2014 \`dry_run\` first)" | leadbay_qualify_leads(prior_deliveries: {job_id}) |
 | Delivered leads look right | "Draft outreach for the top ones" | leadbay_prepare_outreach |
 | Delivered 0 or off-profile | "Reshape the example and retry" (name the fix from funnel + scope_notes) | leadbay_find_new_leads (NEW request_id) |
-| Stopped at cost cap (\`stop_reason: max_cost\`) | "Raise the cap to X and get the remaining N" \u2014 X in the account's currency per the funnel-line rule, never a hard-coded \`$\` | leadbay_find_new_leads, NEW request_id (same-id only dedupes onto a LIVE job) + higher max_cost + \`count\` = the SHORTFALL (\`items_requested\` \u2212 delivered), not the original + \`exclude_lead_ids\` = the examined-but-REJECTED ids (novelty covers delivered; these are what it misses \u2014 without them the rerun re-buys the same losers) |
+| Stopped at the job's usage cap (\`stop_reason: max_cost\`) | "Raise the job's cap and get the remaining N" \u2014 no amount, no currency | leadbay_find_new_leads, NEW request_id (same-id only dedupes onto a LIVE job) + higher max_cost + \`count\` = the SHORTFALL (\`items_requested\` \u2212 delivered), not the original + \`exclude_lead_ids\` = the examined-but-REJECTED ids (novelty covers delivered; these are what it misses \u2014 without them the rerun re-buys the same losers) |
 | Stopped on org quota (\`stop_reason: quota\`) | "Check which window is exhausted and when it resets" \u2014 never a re-run: it cannot clear an org quota and burns a submit slot to stop in the same place | leadbay_account_status |
 | Stopped on org quota and the user does not want to wait | "Top up to finish this run" | leadbay_create_topup_link |
 | User wants these tracked in Leadbay | "Add the keepers to a campaign" | leadbay_create_campaign / leadbay_add_leads_to_campaign |
@@ -6433,7 +6441,7 @@ Per step: \`gate_label\` / \`gate_description\` are the widget's forward option,
 | 1 | Check my account | \`leadbay_account_status\` (no args) |
 | 2 | Pull today's leads | \`leadbay_pull_leads\` (no args) |
 | 3 | Draft the first email | \`leadbay_prepare_outreach\` \u2014 \`leadId\` ONLY, never \`enrich\` |
-| 4 | Find who to email | \`leadbay_enrich_titles\` \u2014 free preview, then a consented paid reveal |
+| 4 | Find who to email | \`leadbay_enrich_titles\` \u2014 free preview, then a reveal on consent |
 
 Steps 1, 2 and 3 carry \`branches[]\`, and steps 3 and 4 carry \`spend\` (+ \`quota_note\` on 4). Every step also carries \`explain\` (say this BEFORE firing) and \`next_steps\` (\`{question, options[]}\` \u2014 already the widget's shape, map it verbatim).
 
@@ -6445,9 +6453,9 @@ The manifest also carries **\`keep_going\`**: the closing cheat-sheet of *what y
 
 **Step 1 shows the real account, and is silent about two things.** The click is labelled *check my account status*, so deliver it: user + org, then the **full quota windows** the way the web app renders them \u2014 Daily / Weekly / Monthly with a \`\u25B0\u25B1\` gauge, % used, $ spent against the cap, resets countdown, and the per-resource breakdown. Never raw "credits". But apply the silence gate first: when \`quota\` is null, \`quota_error\` is set, or the org has \`unlimited_credits\`, say **nothing** about quota \u2014 never mention a 401, never suggest logging in again (the token is fine, the same response just read their account), and never announce "unlimited". And **never volunteer the lens**: the response withholds it unless the user asked, so there is nothing to report and no other tool to reach for. Both are pinned regressions (WORKFLOWS #30 / #31).
 
-**Step 3 drafts, and spends nothing.** Call \`leadbay_prepare_outreach\` with \`leadId\` alone \u2014 **never \`enrich: true\`**, which launches a paid contact reveal off the back of a *draft* click. \`recommended_contact\` returns with \`email\`/\`phone\` null; that is expected, and it is the hook for step 4. Render through \`message_compose_v1\` (2\u20133 strategy-labelled variants), address it to the job TITLE \u2014 no name exists yet, and inventing one is fabrication \u2014 and never send it or offer to.
+**Step 3 drafts, and uses no quota.** Call \`leadbay_prepare_outreach\` with \`leadId\` alone \u2014 **never \`enrich: true\`**, which launches a contact reveal off the back of a *draft* click. \`recommended_contact\` returns with \`email\`/\`phone\` null; that is expected, and it is the hook for step 4. Render through \`message_compose_v1\` (2\u20133 strategy-labelled variants), address it to the job TITLE \u2014 no name exists yet, and inventing one is fabrication \u2014 and never send it or offer to.
 
-**Step 4 runs in two beats \u2014 free first, paid only on consent.** Scoped to the ONE lead step 3 drafted for. Beat 1 omits \`titles\` and returns \`mode:"discover"\`, the free list of job titles at that company; say plainly that nothing has been spent. Beat 2 names the title the draft is addressed to, states the cost BEFORE they decide (one contact, one credit), and only on confirmation calls again with \`titles\` + \`confirm:true\` + \`email:true\` \u2014 polled via \`leadbay_bulk_enrich_status\` until done, reporting only what actually resolved. The gate click bought the free look, not the reveal: never launch without an explicit confirm.
+**Step 4 runs in two beats \u2014 free first, the reveal only on consent.** Scoped to the ONE lead step 3 drafted for. Beat 1 omits \`titles\` and returns \`mode:"discover"\`, the free list of job titles at that company; say plainly that nothing has run yet. Beat 2 names the title the draft is addressed to, says BEFORE they decide that revealing one contact uses a little of their plan's quota (no amount, no price), and only on confirmation calls again with \`titles\` + \`confirm:true\` + \`email:true\` \u2014 polled via \`leadbay_bulk_enrich_status\` until done, reporting only what actually resolved. The gate click bought the free look, not the reveal: never launch without an explicit confirm.
 
 ## Empty first batch is normal, not an error
 
@@ -6484,7 +6492,7 @@ WHEN NOT TO USE: discovery (use leadbay_pull_leads); single-lead deep dive (use
 Leadbay has no cancel. Once \`leadbay_enrich_titles\`, \`leadbay_bulk_qualify_leads\`,
 \`leadbay_import_leads\` or \`leadbay_import_and_qualify\` has returned a launched or
 running result, that work is queued on Leadbay and runs to completion, and the
-quota it costs is already committed. A discovery, preview or \`dry_run\` result
+quota it uses is already committed. A discovery, preview or \`dry_run\` result
 launched nothing and is not covered here.
 
 The user cancelling in the chat, a request timeout, or a closed stream stops YOUR
@@ -6492,7 +6500,7 @@ waiting, never the job. \`cancelled: true\` means we stopped watching, not that
 work stopped. What to do next depends on what you are holding:
 
 - **A handle.** Poll the status tool with it, and do not launch the work that
-  handle covers a second time \u2014 that spends the quota again on the same rows.
+  handle covers a second time \u2014 that uses the quota again on the same rows.
   \`leadbay_import_status\` takes \`importIds\`, so pass the values of \`import_ids\`
   under that name. A qualification started by \`leadbay_import_and_qualify\` has no
   notification of its own: resume it with
@@ -6506,7 +6514,7 @@ work stopped. What to do next depends on what you are holding:
   finished. Calling the same tool again with the same arguments will usually hand
   back the job already launched rather than starting a second one, but that guard
   is in-memory, five minutes, and per process, so it is best-effort \u2014 say what you
-  are about to re-run before you spend the user's quota on it.
+  are about to re-run before you use the user's quota on it.
 
 
 Budgets: \`total_budget_ms\` caps wall-clock; \`per_lead_budget_ms\` caps each lead's poll. For short transport timeouts, pass \`wait_for_completion:false\` and poll \`leadbay_import_status\`. Outputs \`qualified[]\`, \`still_running[]\`, \`not_imported[]\`, plus the ids that resume it: \`lead_ids\` + \`lens_id\` for leadbay_qualify_status, \`import_ids\` for leadbay_import_status. There is no qualification \`notification_id\` \u2014 the qualify phase runs per-lead, so no job notification exists; \`notification_ids[]\` are the file-import ones. Idempotent within a 5-min window. \`dry_run:'preview'\` returns mapping hints + custom-field candidates without importing.
@@ -6729,7 +6737,7 @@ a launcher only for a subset that never started, never for the whole batch:
   that are the ones that never started. Only then call
   \`leadbay_bulk_qualify_leads({leadIds, lensId})\` for exactly those ids. A lead
   that is merely slow looks identical to one that never launched over a few
-  polls, and re-launching it charges the user twice \u2014 when unsure, tell the user
+  polls, and re-launching it uses the user's quota twice \u2014 when unsure, tell the user
   rather than guess.
 
 
@@ -6796,18 +6804,18 @@ WHEN NOT TO USE: from agent flow \u2014 leadbay_enrich_titles handles selection
 ## A launched job cannot be stopped, and this tool has no retry guard
 
 Leadbay has no cancel. Once this call returns having actually launched, the work
-is queued on Leadbay and runs to completion, and the quota it costs is already
-committed. A \`dry_run\` result reached no backend and spent nothing. The user
+is queued on Leadbay and runs to completion, and the quota it uses is already
+committed. A \`dry_run\` result reached no backend and used nothing. The user
 cancelling in the chat, a request timeout, or a closed stream stops YOUR waiting,
 never the job.
 
 Unlike the composite launchers, this tool has **no double-launch guard**: calling
-it again always issues a new paid launch, even seconds later with identical
+it again always issues a new launch that uses quota again, even seconds later with identical
 arguments. So when a call returns nothing at all, do not simply retry. Read the
 record back first \u2014 \`leadbay_research_lead_by_id\` or \`leadbay_get_contacts\` for a
 lead, \`leadbay_account_status\` for background work that has since finished \u2014 to
 see whether the launch already landed, and tell the user what you are about to
-spend before spending it again.
+run before running it again.
 
 
 This tool MUTATES state. The caller (agent or human-in-the-loop) is responsible for confirming intent before invocation; the MCP server does not soft-prompt for confirmation. See \`annotations.destructiveHint\`.
@@ -6832,14 +6840,16 @@ Examples that should NOT invoke this tool (sound similar, route elsewhere):
 
 Terminal job -> render the full delivery per the lead-delivery table +
 honest funnel line. Still running -> one progress line (examined /
-delivered / spent so far) and offer to check again in ~1 min. Never
+delivered so far) and offer to check again in ~1 min. Never
 render UUIDs or cursors.
 
 ---
 
 Cumulative snapshot of a lead-delivery job: state, funnel counters, every
-item emitted so far (full lead payloads for delivered/degraded, honest
-status_reason for skipped), spend + breakdown, and the \`explain\` block
+item emitted so far (compact lead payloads for delivered/degraded, with
+one reason and the freshest signals; full evidence via
+\`leadbay_research_lead_by_id\`; honest status_reason for skipped), internal
+usage counters (never render them), and the \`explain\` block
 (basis, seed strategy, scope notes). Items are immutable once emitted \u2014
 polling never re-reads live data, so numbers only ever grow.
 
@@ -6848,7 +6858,7 @@ completion when the user asked for results "in this reply". \`since\` (from a
 prior poll's \`next_since\`) pages only the new items. Jobs terminalize
 server-side: past the 30-min wall clock a job reads \`completed_partial\`
 (time budget), past 30 days \`expired\` (items no longer listed \u2014 re-read
-billed leads via leadbay_qualify_leads \`prior_deliveries\`). A 404 means
+delivered leads via leadbay_qualify_leads \`prior_deliveries\`). A 404 means
 unknown job or another org's job.
 
 ---
@@ -6883,7 +6893,7 @@ when nothing was delivered.
 **Column 3 \u2014 Contact**
 
 - \`[Name](linkedin) \xB7 role\` (linked name mandatory when a LinkedIn URL
-  exists; plain name otherwise). Below it, the PURCHASED channels only:
+  exists; plain name otherwise). Below it, the FOUND channels only:
   \`\u2709 value\` / \`\u260E value\` inline as plain text (they auto-linkify).
 - Channel statuses: \`delivered\` \u2192 show value; \`already_owned\` \u2192 value +
   *(already yours)*; \`masked\` \u2192 "on file \u2014 reveal via channels";
@@ -6894,16 +6904,17 @@ when nothing was delivered.
 
 **The funnel line (mandatory, after the table):**
 
-One short line narrating the delivery honestly, from \`funnel\` + \`cost\` +
+One short line narrating the delivery honestly, from \`funnel\` +
 \`explain.scope_notes\`:
 
 > Matched N \xB7 examined E \xB7 qualified Q \xB7 disqualified D \u2192 **delivered X of
-> the Y asked** \xB7 stopped: <stop_reason in plain words> \xB7 spent C.CC.
+> the Y asked** \xB7 stopped: <stop_reason in plain words>.
 
-**Money: divide, then symbol.** Every amount (\`cost.spent\`,
-\`estimated_cost.max\`, quotes) is \`cost_cents\` \u2014 divide by 100, two decimals,
-so \`165\` renders \`1.65\`, NEVER \`165.00\`. Symbol from the account region: US
-\`$\`, France \`\u20AC\`, unknown \u2192 bare. Never hard-code \`$\`: it misstates a charge.
+**No money, anywhere.** \`cost.*\`, \`estimated_cost.max\` and quotes are internal
+usage units. Never render them, never convert them to a currency, never call
+them a charge: the user's plan or top-up covers this work, and a price reads as
+a bill. If the user asks what a job used, show \`leadbay_account_status\`'s quota
+windows.
 
 "of the Y asked" needs \`summary.items_requested\`, which submits carry but a
 later \`leadbay_lead_job_status\` snapshot does not. Without it write **delivered
@@ -6911,7 +6922,7 @@ X** and stop \u2014 never back-fill Y from \`matched\`/\`examined\` (they count
 candidates), never guess it.
 
 Plain-word stop reasons: \`target_reached\` \u2192 omit (success), \`pool_exhausted\` \u2192
-"ran out of matching candidates", \`max_cost\` \u2192 "hit the cost cap", \`quota\` \u2192
+"ran out of matching candidates", \`max_cost\` \u2192 "hit the job's usage cap", \`quota\` \u2192
 "hit an org quota", \`time_budget\` \u2192 "hit the 30-min time budget".
 
 **When \`delivered\` is 0**: NEVER say just "no results". Render no table; give
@@ -6936,7 +6947,7 @@ table \`Ref \u2192 Outcome\` translating \`status_reason\` to plain words:
 \`low_confidence_identity\` \u2192 "couldn't safely match \u2014 check \`resolution.alternatives\`",
 \`no_matching_contact\` \u2192 "no contact with the requested title",
 \`disqualified\` \u2192 "evaluated: does not fit" (evidence is in the item when owned),
-\`enrichment_failed\` \u2192 "channel could not be sourced (not billed)".
+\`enrichment_failed\` \u2192 "channel could not be sourced (no quota used)".
 
 **\`items_truncated\`**: rows are a PREFIX, not the batch. Say so, and offer
 \`leadbay_lead_job_status(job_id, since: next_since)\` for the rest.
@@ -7003,7 +7014,7 @@ is a status tool, keep it terse:
 |---|---|---|
 | Still running | "Keep waiting (~1 min) or leave it \u2014 results are kept 30 days" | leadbay_lead_job_status(job_id, wait_seconds: 60) |
 | Terminal (completed / partial / failed) | Render the delivery per the RENDERING block, then offer the matching find_new_leads / qualify_leads NEXT STEPS | \u2014 |
-| \`expired\` (past the 30-day window) | "Re-read the billed leads from your delivery ledger" \u2014 there is nothing left to render: the job terminalized and its items are no longer listed, so do NOT present an empty delivery as a result | leadbay_qualify_leads(prior_deliveries: {job_id}) |
+| \`expired\` (past the 30-day window) | "Re-read the delivered leads from your delivery ledger" \u2014 there is nothing left to render: the job terminalized and its items are no longer listed, so do NOT present an empty delivery as a result | leadbay_qualify_leads(prior_deliveries: {job_id}) |
 `;
 var leadbay_like_lead = `## WHEN TO USE
 
@@ -7679,9 +7690,9 @@ Offer 2\u20133 follow-ups. Choose based on enrichment state + available channels
 | After a successful exchange                     | "Update qualification answers based on what you learned"      | leadbay_answer_clarification                           |
 The "log outreach" step is the most-important follow-up \u2014 it closes the loop and populates history for the next \`leadbay_prepare_outreach\` call. Detect intent from natural language: "I sent the email", "she didn't pick up", "left a voicemail", "they responded yes/no", etc.
 `;
-var leadbay_preview_bulk_enrichment = `Preview a bulk-enrichment cost given a set of job titles applied to the current selection. Returns \`{selected_leads, enriched_contacts, enrichable_contacts, title_suggestions, auto_included_titles, previously_enriched_titles}\`. \`previously_enriched_titles\` is a newer field (in prod soon) \u2014 when present, the agent can recommend repeating those titles for new leads.
+var leadbay_preview_bulk_enrichment = `Preview a bulk enrichment given a set of job titles applied to the current selection. Returns \`{selected_leads, enriched_contacts, enrichable_contacts, title_suggestions, auto_included_titles, previously_enriched_titles}\`. \`previously_enriched_titles\` is a newer field (in prod soon) \u2014 when present, the agent can recommend repeating those titles for new leads.
 
-WHEN TO USE: between selecting leads and launching, to know what the enrichment will cost.
+WHEN TO USE: between selecting leads and launching, to know how many contacts the enrichment will reveal.
 
 WHEN NOT TO USE: from agent flow \u2014 leadbay_enrich_titles wraps preview + launch with the right safety checks.
 `;
@@ -8076,18 +8087,18 @@ WHEN NOT TO USE: as the agent's bulk-qualify path \u2014 use leadbay_bulk_qualif
 ## A launched job cannot be stopped, and this tool has no retry guard
 
 Leadbay has no cancel. Once this call returns having actually launched, the work
-is queued on Leadbay and runs to completion, and the quota it costs is already
-committed. A \`dry_run\` result reached no backend and spent nothing. The user
+is queued on Leadbay and runs to completion, and the quota it uses is already
+committed. A \`dry_run\` result reached no backend and used nothing. The user
 cancelling in the chat, a request timeout, or a closed stream stops YOUR waiting,
 never the job.
 
 Unlike the composite launchers, this tool has **no double-launch guard**: calling
-it again always issues a new paid launch, even seconds later with identical
+it again always issues a new launch that uses quota again, even seconds later with identical
 arguments. So when a call returns nothing at all, do not simply retry. Read the
 record back first \u2014 \`leadbay_research_lead_by_id\` or \`leadbay_get_contacts\` for a
 lead, \`leadbay_account_status\` for background work that has since finished \u2014 to
 see whether the launch already landed, and tell the user what you are about to
-spend before spending it again.
+run before running it again.
 
 
 This tool MUTATES state. The caller (agent or human-in-the-loop) is responsible for confirming intent before invocation; the MCP server does not soft-prompt for confirmation. See \`annotations.destructiveHint\`.
@@ -8115,7 +8126,7 @@ Examples that should NOT invoke this tool (sound similar, route elsewhere):
 3-col table for delivered items (fit bar + company / why-fits \u226420 words /
 contact + channels) in returned order, then a compact Ref \u2192 Outcome table
 for skipped refs (not_in_universe, low_confidence_identity, ... in plain
-words), then the honest funnel + cost line. Full algorithm below.
+words), then the honest funnel line. Full algorithm below.
 
 ---
 
@@ -8130,29 +8141,29 @@ job; this tool polls up to \`wait_seconds\` (default 45) and hands off to
 **Refs are flexible; outcomes are per-item.** \`lead_refs\` accepts any mix of
 \`lead_id\`, \`website\`, \`name\`(+\`location\`), or a stable \`contact_id\` from a
 prior result (enrichment then targets exactly that person, never a re-match).
-\`prior_deliveries\` expands past MCP deliveries into refs \u2014 billed leads stay
+\`prior_deliveries\` expands past MCP deliveries into refs \u2014 delivered leads stay
 re-readable this way even after the 30-day result window. Duplicates collapse.
 A ref that can't be served comes back \`skipped\` with an honest
 \`status_reason\` (\`not_in_universe\`, \`low_confidence_identity\` with the
 \`resolution.alternatives\` to choose from, \`no_matching_contact\`, ...) \u2014 that
-is an ANSWER about the ref, not an error, and it costs nothing.
+is an ANSWER about the ref, not an error, and it uses no quota.
 
 **Disqualified \u2260 dropped.** Companies the org owns that fail qualification
 are DELIVERED with their negative evidence (question verdicts, tag misses,
 IBP reasoning) \u2014 "here's why to skip this account" is a deliverable.
 
-**Cost \u2014 never spend silently.** Resolution and identity are free.
-\`qualify: true\` (the default) costs ~94 cost_cents per lead needing FRESH
-research+scoring \u2014 but repeat calls reuse every fresh cached stage
-(\`from_cache\` flags on the items) and converge to near-zero cost. \`channels\`
-purchase verified email (25c) / phone (250c) on success only;
-\`already_owned\` values cost nothing.
+**Usage quota \u2014 never use it silently.** Resolution and identity are free.
+\`qualify: true\` (the default) draws on the org's usage quota for each lead
+needing FRESH research+scoring \u2014 but repeat calls reuse every fresh cached stage
+(\`from_cache\` flags on the items) and converge to near-zero. \`channels\` draw on
+it only when a verified email / phone is found; \`already_owned\` values are
+free. Nothing is invoiced.
 
-The gate is enforced in code, not just here: a PAID call (\`qualify\` left at
+The gate is enforced in code, not just here: a call that uses quota (\`qualify\` left at
 its default or set true, and/or any \`channels\`) is WITHHELD unless it carries
 \`confirm: true\`. Without it the tool submits nothing and returns
 \`mode: "needs_confirmation"\` with a real backend quote \u2014 show that quote to
-the user, get the go-ahead (an explicit "spend / get their emails" in their
+the user, get the go-ahead (an explicit "go ahead / get their emails" in their
 message counts), then re-call with \`confirm: true\`. \`confirm: false\` is a
 veto: nothing is submitted and no quote round-trip is made. A fully FREE
 call (\`qualify: false\`, no \`channels\`) needs no \`confirm\` and passes straight
@@ -8193,7 +8204,7 @@ when nothing was delivered.
 **Column 3 \u2014 Contact**
 
 - \`[Name](linkedin) \xB7 role\` (linked name mandatory when a LinkedIn URL
-  exists; plain name otherwise). Below it, the PURCHASED channels only:
+  exists; plain name otherwise). Below it, the FOUND channels only:
   \`\u2709 value\` / \`\u260E value\` inline as plain text (they auto-linkify).
 - Channel statuses: \`delivered\` \u2192 show value; \`already_owned\` \u2192 value +
   *(already yours)*; \`masked\` \u2192 "on file \u2014 reveal via channels";
@@ -8204,16 +8215,17 @@ when nothing was delivered.
 
 **The funnel line (mandatory, after the table):**
 
-One short line narrating the delivery honestly, from \`funnel\` + \`cost\` +
+One short line narrating the delivery honestly, from \`funnel\` +
 \`explain.scope_notes\`:
 
 > Matched N \xB7 examined E \xB7 qualified Q \xB7 disqualified D \u2192 **delivered X of
-> the Y asked** \xB7 stopped: <stop_reason in plain words> \xB7 spent C.CC.
+> the Y asked** \xB7 stopped: <stop_reason in plain words>.
 
-**Money: divide, then symbol.** Every amount (\`cost.spent\`,
-\`estimated_cost.max\`, quotes) is \`cost_cents\` \u2014 divide by 100, two decimals,
-so \`165\` renders \`1.65\`, NEVER \`165.00\`. Symbol from the account region: US
-\`$\`, France \`\u20AC\`, unknown \u2192 bare. Never hard-code \`$\`: it misstates a charge.
+**No money, anywhere.** \`cost.*\`, \`estimated_cost.max\` and quotes are internal
+usage units. Never render them, never convert them to a currency, never call
+them a charge: the user's plan or top-up covers this work, and a price reads as
+a bill. If the user asks what a job used, show \`leadbay_account_status\`'s quota
+windows.
 
 "of the Y asked" needs \`summary.items_requested\`, which submits carry but a
 later \`leadbay_lead_job_status\` snapshot does not. Without it write **delivered
@@ -8221,7 +8233,7 @@ X** and stop \u2014 never back-fill Y from \`matched\`/\`examined\` (they count
 candidates), never guess it.
 
 Plain-word stop reasons: \`target_reached\` \u2192 omit (success), \`pool_exhausted\` \u2192
-"ran out of matching candidates", \`max_cost\` \u2192 "hit the cost cap", \`quota\` \u2192
+"ran out of matching candidates", \`max_cost\` \u2192 "hit the job's usage cap", \`quota\` \u2192
 "hit an org quota", \`time_budget\` \u2192 "hit the 30-min time budget".
 
 **When \`delivered\` is 0**: NEVER say just "no results". Render no table; give
@@ -8246,7 +8258,7 @@ table \`Ref \u2192 Outcome\` translating \`status_reason\` to plain words:
 \`low_confidence_identity\` \u2192 "couldn't safely match \u2014 check \`resolution.alternatives\`",
 \`no_matching_contact\` \u2192 "no contact with the requested title",
 \`disqualified\` \u2192 "evaluated: does not fit" (evidence is in the item when owned),
-\`enrichment_failed\` \u2192 "channel could not be sourced (not billed)".
+\`enrichment_failed\` \u2192 "channel could not be sourced (no quota used)".
 
 **\`items_truncated\`**: rows are a PREFIX, not the batch. Say so, and offer
 \`leadbay_lead_job_status(job_id, since: next_since)\` for the rest.
@@ -8314,7 +8326,7 @@ Pick the 2-3 options that match what actually happened:
 | Fit leads with contacts delivered | "Draft outreach for the qualified ones" | leadbay_prepare_outreach |
 | Items skipped \`not_in_universe\` | "Import those companies first, then re-qualify" | leadbay_import_leads \u2192 leadbay_qualify_leads |
 | Items skipped \`low_confidence_identity\` | "Pick the right match" (show \`resolution.alternatives\`) | leadbay_qualify_leads with the chosen lead_id |
-| Contacts delivered without channels | "Purchase verified emails/phones for the keepers (state cost first)" | leadbay_qualify_leads(lead_refs with contact_id, channels) |
+| Contacts delivered without channels | "Get verified emails/phones for the keepers (uses quota \u2014 say so first)" | leadbay_qualify_leads(lead_refs with contact_id, channels) |
 | Disqualified with evidence | "Review why \u2014 adjust qualification questions if the criteria are off" | leadbay_get_qualification_questions |
 `;
 var leadbay_qualify_status = `Retrieve the current state of a bulk_qualify_leads or import_and_qualify launch. Which ids to pass depends on which tool launched it, because only one of them creates a qualification job on the backend:
@@ -8360,7 +8372,7 @@ a launcher only for a subset that never started, never for the whole batch:
   that are the ones that never started. Only then call
   \`leadbay_bulk_qualify_leads({leadIds, lensId})\` for exactly those ids. A lead
   that is merely slow looks identical to one that never launched over a few
-  polls, and re-launching it charges the user twice \u2014 when unsure, tell the user
+  polls, and re-launching it uses the user's quota twice \u2014 when unsure, tell the user
   rather than guess.
 `;
 var leadbay_recall_ordered_titles = `Show job titles the org has previously enriched, so the agent can repeat the same titles for new leads (or skip already-saturated ones). Two implementation paths: (1) PREFERRED \u2014 a selection-scoped preview call that reads \`previously_enriched_titles\` from the backend (newer prod field). (2) FALLBACK \u2014 live aggregation across each lead's enriched contacts. The composite picks transparently.
@@ -9912,9 +9924,9 @@ Examples that should NOT invoke this tool (sound similar, route elsewhere):
 
 ## RENDER (quick)
 
-Report user + org, AND quota whenever readable \u2014 include quota even on a plain
-"what account?" ask. NEVER mention the lens unless asked (use
-\`last_requested_lens_name\`, never the id). SILENT on quota ONLY when
+Report user + org; show quota only when asked (quota / account status) or a
+window is exhausted. NEVER mention the lens unless asked (use
+\`last_requested_lens_name\`, never the id). SILENT on quota when
 \`quota_error\` set, \`unlimited_credits\` true, or quota null. Else render
 Daily/Weekly/Monthly from \`quota.user\` (fall back to \`quota.org\` if \`user\`
 absent) as \`$used / $cap (N% used) \xB7 resets\` (or a resource-count table when
@@ -9922,7 +9934,7 @@ absent) as \`$used / $cap (N% used) \xB7 resets\` (or a resource-count table whe
 
 ---
 
-Show the user's account state \u2014 admin rights, language, last-active lens, quota usage across daily/weekly/monthly windows, and whether the org's intelligence is mid-regeneration. **Show quota the way the web app does \u2014 a percentage-used + dollar-spend gauge per window, never raw "credits".** Each window in \`quota.<group>.spend[]\` carries \`current_units\` / \`max_units\` in dollar_cents (% used = the ratio, $ = \`/100\`); the \`quota.<group>.resources[]\` list gives the per-resource usage breakdown (\`count\`, plus \`max_units\` when a per-resource cap exists). **Pre-check the \`LENS_EXTRA_REFILL\` resource here before calling \`leadbay_extend_lens\`** \u2014 look in **\`quota.org.resources[]\`** first (admins), and fall back to **\`quota.user.resources[]\`** when \`quota.org\` is absent (non-admin callers only get the \`user\` group), matching the resource type **case-insensitively** (it may arrive as \`LENS_EXTRA_REFILL\` or \`lens_extra_refill\`). Its full requested batch must fit into the remaining daily quota or the call is rejected outright. Quota windows also hint at the user's consumption pace: heavy recent activity (ai_rescore / web_fetch near their window limits) is a signal that Leadbay will deliver a larger fresh batch next time the user logs back in, since batch size is paced by real consumption.
+Show the user's account state \u2014 admin rights, language, last-active lens, quota usage across daily/weekly/monthly windows, and whether the org's intelligence is mid-regeneration. **When you show quota, show it the way the web app does \u2014 a percentage-used + dollar-spend gauge per window, never raw "credits".** Each window in \`quota.<group>.spend[]\` carries \`current_units\` / \`max_units\` in dollar_cents (% used = the ratio, $ = \`/100\`); the \`quota.<group>.resources[]\` list gives the per-resource usage breakdown (\`count\`, plus \`max_units\` when a per-resource cap exists). **Pre-check the \`LENS_EXTRA_REFILL\` resource here before calling \`leadbay_extend_lens\`** \u2014 look in **\`quota.org.resources[]\`** first (admins), and fall back to **\`quota.user.resources[]\`** when \`quota.org\` is absent (non-admin callers only get the \`user\` group), matching the resource type **case-insensitively** (it may arrive as \`LENS_EXTRA_REFILL\` or \`lens_extra_refill\`). Its full requested batch must fit into the remaining daily quota or the call is rejected outright. Quota windows also hint at the user's consumption pace: heavy recent activity (ai_rescore / web_fetch near their window limits) is a signal that Leadbay will deliver a larger fresh batch next time the user logs back in, since batch size is paced by real consumption.
 
 **After a user tops up, do NOT keep refusing \u2014 RETRY.** If the user signals they topped up / bought credits / added credits, the previous QUOTA_EXCEEDED is invalidated the moment the Stripe webhook lands. RE-CALL \`leadbay_account_status\` to pick up the new state AND retry the originally failed call. The retry itself does not require a successful account_status check first \u2014 a topped-up user has cleared the throttle whether or not your cached snapshot reflects it yet. **A stale quota snapshot is never a reason to gate-keep a topped-up user.**
 
@@ -9956,7 +9968,7 @@ Some Leadbay tool responses include a \`_meta.notifications\` array listing **ba
 
 - **Previous turn / before an MCP restart, and the user has NOT asked about it** \u2014 don't poll for it in the background. Simply continue the conversation; the next time you call any tool, the completed-work entry appears in \`_meta.notifications\` (also on \`leadbay_account_status.notifications\`). This is the ambient push path \u2014 leave it to do its job. **But if the user explicitly asks for status or to "wait for it to finish"** (e.g. a multi-turn flow where a job was launched in a prior turn and this turn says "wait for enrichment to finish, then \u2026"), DO poll its status tool now until done, exactly as for a this-turn job below \u2014 the ambient push only surfaces *completed* work, so it can't answer a live "is it done / wait for it" request while the job is still running.
 - **This turn (you just launched it)** \u2014 the DEFAULT is: do NOT end your turn on the "launched" ack; stay active and poll the job's status tool in a loop until it reports done, then report the finished result yourself, rather than spinning forever or deferring the result to a later turn. (Two exceptions, detailed below: the user explicitly asked NOT to wait / to run it in the background; or it's a large qualification/import that's async by design \u2014 in those cases hand back the handle instead of looping.) Each status tool has its OWN terminal signal \u2014 poll until:
-  - \`leadbay_bulk_enrich_status\` \u2192 \`all_done:true\` \u2014 OR \`overall_progress.done\` holds steady across several SPACED polls (~15\u201330s apart) over at least ~90s\u20132 min of elapsed time (some contacts are unresolvable, so \`all_done\` can stay false forever). Don't call a plateau from the first few back-to-back reads \u2014 early on \`done\` sits flat while the backend spins up. Once the plateau is real, report what resolved and name what didn't.
+  - \`leadbay_bulk_enrich_status\` \u2192 \`all_done:true\` \u2014 OR \`overall_progress.done\` holds steady across several SPACED polls (~15\u201330s apart) over at least ~90s\u20132 min of elapsed time (a reservation the provider never answers stays \`done:false\`, so \`all_done\` can stay false forever). Don't call a plateau from the first few back-to-back reads \u2014 early on \`done\` sits flat while the backend spins up. Once the plateau is real, report what resolved and name what didn't.
   - \`leadbay_qualify_status\` \u2192 \`still_running\` is empty: every launched lead has finished or failed. (\`in_progress\` also reads \`false\` on the fast path, but it can be \`null\` on the legacy/fallback read \u2014 so treat an empty \`still_running\` as terminal on its own; only require \`in_progress:false\` when that field is actually present.) LIKE imports, large qualification runs are async by design: \`leadbay_bulk_qualify_leads\` defaults to \`wait_for_completion:false\` for \`count > 5\` or chained workflows because blocking can time out, and \`leadbay_qualify_status\` may take minutes/hours. So don't force a long polling loop on a big run \u2014 return the handle/progress and let completion arrive via \`_meta.notifications\` \u2014 UNLESS the user explicitly asked to wait, or it's a small run that finishes quickly. A small \`wait_for_completion:true\` run you can poll to \`still_running\` empty inline.
   - \`leadbay_import_status\` \u2192 \`status:"complete"\` (or \`"failed"\`). BUT imports are the exception to the stay-active loop: a large \`leadbay_import_leads({wait_for_completion:false})\` is meant to return a handle and resolve over minutes, and the tool does ONE refresh pass per call. Don't block the conversation looping on it \u2014 surface the returned progress/handle and let the completion arrive via \`_meta.notifications\` \u2014 UNLESS the user explicitly asked you to wait for the import, or it's a small import that finishes quickly.
 
@@ -9974,10 +9986,10 @@ Mirror the Leadbay web quota widget: three windows side by side \u2014 **Daily**
 $ cap** figure, with a per-resource usage breakdown underneath. **Never speak in
 raw "credits"** for quota \u2014 the unit is a percentage and a dollar spend.
 
-**Include the quota whenever it is readable** \u2014 as part of the default account
-answer, even when the user only asked "what account am I connected to?". The
-sole reason to omit it is the silence gate below (unreadable quota, or an
-unlimited account); it is NOT gated on the user explicitly asking for quota.
+**Show the quota only when it matters** \u2014 when the user asks about their quota,
+usage or account status, or when a window is exhausted and blocks what they
+asked for. A plain "what account am I connected to?" is answered with user +
+org alone. Even then, the silence gate below comes first.
 
 **Silence gate (check FIRST).** Render NOTHING about quota when any of these
 holds \u2014 do not mention quota at all, do not say "unreadable", never tell the user
@@ -10037,7 +10049,7 @@ days"), computed against now \u2014 mirroring the widget's "r\xE9initialis\xE9 d
 raw value is an ISO-8601 timestamp.
 
 **Top-up (optional, subordinate).** When \`quota.topup\` is present, you MAY add one
-small line below the windows: \`Top-up: $<remaining_cents/100> of $<total_credit_cents/100> left\`.
+small line below the windows: \`Top-up: $<(remaining_cents / 100).toFixed(2)> of $<(total_credit_cents / 100).toFixed(2)> left\`.
 Keep it secondary \u2014 the three window gauges are the headline. Omit when null.
 
 **Legend** (once, below): \`\` \`\u25B0\` used \xB7 \`\u25B1\` remaining \`\`.
@@ -10045,7 +10057,7 @@ Keep it secondary \u2014 the three window gauges are the headline. Omit when nul
 
 ---
 
-WHEN TO USE: at the start of a session to know what the agent can/can't do, after a 429 to explain to the user which resource window was exhausted and when it resets, and after the user signals a top-up so the agent can resume the interrupted workflow.
+WHEN TO USE: when the user asks about their account, quota or version, after a 429 to explain to the user which resource window was exhausted and when it resets, and after the user signals a top-up so the agent can resume the interrupted workflow.
 
 WHEN NOT TO USE: as a pre-flight gate before bulk ops \u2014 operations themselves return 429; this tool is for context, not gating. And: a recent quota snapshot showing "exhausted" is NOT a reason to refuse a write call when the user has just topped up \u2014 re-call this tool first, then proceed.
 `,
@@ -10071,8 +10083,8 @@ Examples that should NOT invoke this tool (sound similar, route elsewhere):
 
 3-col table of delivered leads in returned order: col 1 = 10-segment fit
 bar + linked company \xB7 location \xB7 size; col 2 = why-fits \u226420 words; col 3
-= contact + purchased channels. ALWAYS close with the honest funnel line
-(matched/examined/delivered/stop reason/spend) \u2014 especially on 0
+= contact + found channels. ALWAYS close with the honest funnel line
+(matched/examined/delivered/stop reason) \u2014 especially on 0
 delivered. Full algorithm below.
 
 ---
@@ -10080,19 +10092,19 @@ delivered. Full algorithm below.
 Submit a net-new lead search: the backend matches an ICP seed against the full
 company universe, applies hard filters, skips what the org already knows
 (\`novelty: org\`), optionally qualifies against the org's own intelligence
-(questions, tags, ideal buyer profile \u2014 frozen at submit), and optionally buys
+(questions, tags, ideal buyer profile \u2014 frozen at submit), and optionally reveals
 contact channels. Polls up to \`wait_seconds\` (default 45); a longer job returns
 \`still_running\` + \`next_poll\` \u2014 hand off to \`leadbay_lead_job_status\`. Jobs run
 \u226430 min, results kept 30 days.
 
-**Free vs paid \u2014 never spend silently.** Default (\`qualify: false\`,
+**Free vs usage quota \u2014 never use quota silently.** Default (\`qualify: false\`,
 \`channels: []\`) is FREE: company profile + fit score + cached research +
-contact identity. Paid: \`qualify: true\` (~94 cost_cents per candidate
-EXAMINED, capped by \`exploration_cap\`/\`max_cost\`) and \`channels\` (email 25c /
-phone 250c, success-only). Enforced in code: a paid call is WITHHELD unless it
+contact identity. \`qualify: true\` (per candidate EXAMINED, capped by
+\`exploration_cap\`/\`max_cost\`) and \`channels\` (only when a value is found) draw
+on the org's usage quota; nothing is invoiced. Enforced in code: such a call is WITHHELD unless it
 carries \`confirm: true\` \u2014 nothing is submitted and you get
 \`mode: "needs_confirmation"\` with a real quote to show the user. Re-call with
-\`confirm: true\` on their go-ahead ("spend / get their emails" counts).
+\`confirm: true\` on their go-ahead ("go ahead / get their emails" counts).
 \`confirm: false\` vetoes. Free needs no consent. **Preview free first** \u2014
 reshaping an off-profile seed is free, exploring it with \`qualify: true\` is
 not.
@@ -10100,7 +10112,7 @@ not.
 **Ad-hoc exclusions ("no chains") are enforced by NO tier** \u2014 \`filters\` has no
 exclusion key, and \`qualify\` scores against the org's FROZEN questions and IBP,
 which need not mention chains; the seed's inverse only shifts ranking.
-Violators can survive, be paid for and be delivered \u2014 post-filter them yourself
+Violators can survive, use quota and be delivered \u2014 post-filter them yourself
 and say the tier didn't enforce it. Durable enforcement \u2192
 \`leadbay_refine_prompt\`.
 
@@ -10147,9 +10159,9 @@ measured:
 
 **Parameter notes**
 - \`request_id\` (REQUIRED) is the retry contract: SAME value retrying the same
-  ask (same live job, no double spend); NEW for a changed ask. Derive from ask
+  ask (same live job, no double launch); NEW for a changed ask. Derive from ask
   + archetype + date: \`gyms-dallas-2026-07-28\`.
-- Never lower \`min_ai_score\` together with \`channels\` \u2014 that buys emails for
+- Never lower \`min_ai_score\` together with \`channels\` \u2014 that reveals emails for
   leads the AI just scored as junk.
 - \`count\` \u2264 50; \u22643 active jobs/org; \u226410 submits/hour (429 + Retry-After \u2014
   wait, don't hammer).
@@ -10189,7 +10201,7 @@ when nothing was delivered.
 **Column 3 \u2014 Contact**
 
 - \`[Name](linkedin) \xB7 role\` (linked name mandatory when a LinkedIn URL
-  exists; plain name otherwise). Below it, the PURCHASED channels only:
+  exists; plain name otherwise). Below it, the FOUND channels only:
   \`\u2709 value\` / \`\u260E value\` inline as plain text (they auto-linkify).
 - Channel statuses: \`delivered\` \u2192 show value; \`already_owned\` \u2192 value +
   *(already yours)*; \`masked\` \u2192 "on file \u2014 reveal via channels";
@@ -10200,16 +10212,17 @@ when nothing was delivered.
 
 **The funnel line (mandatory, after the table):**
 
-One short line narrating the delivery honestly, from \`funnel\` + \`cost\` +
+One short line narrating the delivery honestly, from \`funnel\` +
 \`explain.scope_notes\`:
 
 > Matched N \xB7 examined E \xB7 qualified Q \xB7 disqualified D \u2192 **delivered X of
-> the Y asked** \xB7 stopped: <stop_reason in plain words> \xB7 spent C.CC.
+> the Y asked** \xB7 stopped: <stop_reason in plain words>.
 
-**Money: divide, then symbol.** Every amount (\`cost.spent\`,
-\`estimated_cost.max\`, quotes) is \`cost_cents\` \u2014 divide by 100, two decimals,
-so \`165\` renders \`1.65\`, NEVER \`165.00\`. Symbol from the account region: US
-\`$\`, France \`\u20AC\`, unknown \u2192 bare. Never hard-code \`$\`: it misstates a charge.
+**No money, anywhere.** \`cost.*\`, \`estimated_cost.max\` and quotes are internal
+usage units. Never render them, never convert them to a currency, never call
+them a charge: the user's plan or top-up covers this work, and a price reads as
+a bill. If the user asks what a job used, show \`leadbay_account_status\`'s quota
+windows.
 
 "of the Y asked" needs \`summary.items_requested\`, which submits carry but a
 later \`leadbay_lead_job_status\` snapshot does not. Without it write **delivered
@@ -10217,7 +10230,7 @@ X** and stop \u2014 never back-fill Y from \`matched\`/\`examined\` (they count
 candidates), never guess it.
 
 Plain-word stop reasons: \`target_reached\` \u2192 omit (success), \`pool_exhausted\` \u2192
-"ran out of matching candidates", \`max_cost\` \u2192 "hit the cost cap", \`quota\` \u2192
+"ran out of matching candidates", \`max_cost\` \u2192 "hit the job's usage cap", \`quota\` \u2192
 "hit an org quota", \`time_budget\` \u2192 "hit the 30-min time budget".
 
 **When \`delivered\` is 0**: NEVER say just "no results". Render no table; give
@@ -10242,7 +10255,7 @@ table \`Ref \u2192 Outcome\` translating \`status_reason\` to plain words:
 \`low_confidence_identity\` \u2192 "couldn't safely match \u2014 check \`resolution.alternatives\`",
 \`no_matching_contact\` \u2192 "no contact with the requested title",
 \`disqualified\` \u2192 "evaluated: does not fit" (evidence is in the item when owned),
-\`enrichment_failed\` \u2192 "channel could not be sourced (not billed)".
+\`enrichment_failed\` \u2192 "channel could not be sourced (no quota used)".
 
 **\`items_truncated\`**: rows are a PREFIX, not the batch. Say so, and offer
 \`leadbay_lead_job_status(job_id, since: next_since)\` for the rest.
@@ -10291,10 +10304,10 @@ Pick the 2-3 options that match what actually happened \u2014 never all seven:
 | Observation | Suggest | Calls |
 |---|---|---|
 | Job still running (\`still_running: true\`) | "Check on it in ~1 min" | leadbay_lead_job_status(job_id, wait_seconds: 60) |
-| Free run delivered on-profile leads | "Qualify these N against your criteria (paid \u2014 \`dry_run\` first)" | leadbay_qualify_leads(prior_deliveries: {job_id}) |
+| Free run delivered on-profile leads | "Qualify these N against your criteria (uses quota \u2014 \`dry_run\` first)" | leadbay_qualify_leads(prior_deliveries: {job_id}) |
 | Delivered leads look right | "Draft outreach for the top ones" | leadbay_prepare_outreach |
 | Delivered 0 or off-profile | "Reshape the example and retry" (name the fix from funnel + scope_notes) | leadbay_find_new_leads (NEW request_id) |
-| Stopped at cost cap (\`stop_reason: max_cost\`) | "Raise the cap to X and get the remaining N" \u2014 X in the account's currency per the funnel-line rule, never a hard-coded \`$\` | leadbay_find_new_leads, NEW request_id (same-id only dedupes onto a LIVE job) + higher max_cost + \`count\` = the SHORTFALL (\`items_requested\` \u2212 delivered), not the original + \`exclude_lead_ids\` = the examined-but-REJECTED ids (novelty covers delivered; these are what it misses \u2014 without them the rerun re-buys the same losers) |
+| Stopped at the job's usage cap (\`stop_reason: max_cost\`) | "Raise the job's cap and get the remaining N" \u2014 no amount, no currency | leadbay_find_new_leads, NEW request_id (same-id only dedupes onto a LIVE job) + higher max_cost + \`count\` = the SHORTFALL (\`items_requested\` \u2212 delivered), not the original + \`exclude_lead_ids\` = the examined-but-REJECTED ids (novelty covers delivered; these are what it misses \u2014 without them the rerun re-buys the same losers) |
 | Stopped on org quota (\`stop_reason: quota\`) | "Check which window is exhausted and when it resets" \u2014 never a re-run: it cannot clear an org quota and burns a submit slot to stop in the same place | leadbay_account_status |
 | User wants these tracked in Leadbay | "Add the keepers to a campaign" | leadbay_create_campaign / leadbay_add_leads_to_campaign |
 `,
@@ -10857,7 +10870,7 @@ var getContacts = {
                 },
                 credits_used: {
                   type: "number",
-                  description: "Credits charged for this reveal. Only meaningful when done:true. An explicit 0 alongside done:true means the provider returned nothing. Optional \u2014 when absent the cost is unknown and terminal-empty must NOT be inferred."
+                  description: "Credits used by this reveal. Only meaningful when done:true. An explicit 0 alongside done:true means the provider returned nothing. Optional \u2014 when absent the cost is unknown and terminal-empty must NOT be inferred."
                 },
                 email_requested: { type: "boolean" },
                 phone_requested: { type: "boolean" }
@@ -15383,7 +15396,7 @@ var removePushback = {
 var previewBulkEnrichment = {
   name: "leadbay_preview_bulk_enrichment",
   annotations: {
-    title: "Preview bulk enrichment cost",
+    title: "Preview bulk enrichment",
     readOnlyHint: true,
     destructiveHint: false,
     idempotentHint: true,
@@ -18468,6 +18481,9 @@ var researchLeadById = {
     ]
   },
   execute: async (client, params, _ctx) => {
+    if (typeof params.leadId !== "string" || params.leadId.trim() === "") {
+      throw client.makeError("INVALID_PARAMS", "leadId is required and must be a non-empty string", "Pass the lead's full 36-character UUID as `leadId`, exactly as leadbay_pull_leads items[].id returned it. If you only have the company name, call leadbay_research_lead_by_name_fuzzy instead.");
+    }
     const lensId = params.lensId ?? await client.resolveDefaultLens();
     const leadId = params.leadId;
     void client.request("POST", "/interactions", [
@@ -18647,6 +18663,133 @@ researchLeadById.execute = async (client, params, ctx) => {
   return envelope;
 };
 
+// ../../node_modules/.pnpm/fastest-levenshtein@1.0.16/node_modules/fastest-levenshtein/esm/mod.js
+var peq = new Uint32Array(65536);
+var myers_32 = (a, b) => {
+  const n = a.length;
+  const m = b.length;
+  const lst = 1 << n - 1;
+  let pv = -1;
+  let mv = 0;
+  let sc = n;
+  let i = n;
+  while (i--) {
+    peq[a.charCodeAt(i)] |= 1 << i;
+  }
+  for (i = 0; i < m; i++) {
+    let eq = peq[b.charCodeAt(i)];
+    const xv = eq | mv;
+    eq |= (eq & pv) + pv ^ pv;
+    mv |= ~(eq | pv);
+    pv &= eq;
+    if (mv & lst) {
+      sc++;
+    }
+    if (pv & lst) {
+      sc--;
+    }
+    mv = mv << 1 | 1;
+    pv = pv << 1 | ~(xv | mv);
+    mv &= xv;
+  }
+  i = n;
+  while (i--) {
+    peq[a.charCodeAt(i)] = 0;
+  }
+  return sc;
+};
+var myers_x = (b, a) => {
+  const n = a.length;
+  const m = b.length;
+  const mhc = [];
+  const phc = [];
+  const hsize = Math.ceil(n / 32);
+  const vsize = Math.ceil(m / 32);
+  for (let i = 0; i < hsize; i++) {
+    phc[i] = -1;
+    mhc[i] = 0;
+  }
+  let j = 0;
+  for (; j < vsize - 1; j++) {
+    let mv2 = 0;
+    let pv2 = -1;
+    const start2 = j * 32;
+    const vlen2 = Math.min(32, m) + start2;
+    for (let k = start2; k < vlen2; k++) {
+      peq[b.charCodeAt(k)] |= 1 << k;
+    }
+    for (let i = 0; i < n; i++) {
+      const eq = peq[a.charCodeAt(i)];
+      const pb = phc[i / 32 | 0] >>> i & 1;
+      const mb = mhc[i / 32 | 0] >>> i & 1;
+      const xv = eq | mv2;
+      const xh = ((eq | mb) & pv2) + pv2 ^ pv2 | eq | mb;
+      let ph = mv2 | ~(xh | pv2);
+      let mh = pv2 & xh;
+      if (ph >>> 31 ^ pb) {
+        phc[i / 32 | 0] ^= 1 << i;
+      }
+      if (mh >>> 31 ^ mb) {
+        mhc[i / 32 | 0] ^= 1 << i;
+      }
+      ph = ph << 1 | pb;
+      mh = mh << 1 | mb;
+      pv2 = mh | ~(xv | ph);
+      mv2 = ph & xv;
+    }
+    for (let k = start2; k < vlen2; k++) {
+      peq[b.charCodeAt(k)] = 0;
+    }
+  }
+  let mv = 0;
+  let pv = -1;
+  const start = j * 32;
+  const vlen = Math.min(32, m - start) + start;
+  for (let k = start; k < vlen; k++) {
+    peq[b.charCodeAt(k)] |= 1 << k;
+  }
+  let score = m;
+  for (let i = 0; i < n; i++) {
+    const eq = peq[a.charCodeAt(i)];
+    const pb = phc[i / 32 | 0] >>> i & 1;
+    const mb = mhc[i / 32 | 0] >>> i & 1;
+    const xv = eq | mv;
+    const xh = ((eq | mb) & pv) + pv ^ pv | eq | mb;
+    let ph = mv | ~(xh | pv);
+    let mh = pv & xh;
+    score += ph >>> m - 1 & 1;
+    score -= mh >>> m - 1 & 1;
+    if (ph >>> 31 ^ pb) {
+      phc[i / 32 | 0] ^= 1 << i;
+    }
+    if (mh >>> 31 ^ mb) {
+      mhc[i / 32 | 0] ^= 1 << i;
+    }
+    ph = ph << 1 | pb;
+    mh = mh << 1 | mb;
+    pv = mh | ~(xv | ph);
+    mv = ph & xv;
+  }
+  for (let k = start; k < vlen; k++) {
+    peq[b.charCodeAt(k)] = 0;
+  }
+  return score;
+};
+var distance = (a, b) => {
+  if (a.length < b.length) {
+    const tmp = b;
+    b = a;
+    a = tmp;
+  }
+  if (b.length === 0) {
+    return a.length;
+  }
+  if (a.length <= 32) {
+    return myers_32(a, b);
+  }
+  return myers_x(a, b);
+};
+
 // ../core/dist/composite/research-lead-by-name-fuzzy.js
 var RESOLVE_TIMEOUT_MS = 1e4;
 var MAX_AMBIGUOUS_CANDIDATES = 4;
@@ -18661,6 +18804,28 @@ function suggestionName(suggestion) {
 function suggestionLeadId(suggestion) {
   return suggestion.lead_id ?? suggestion.leadId;
 }
+function nameWords(name) {
+  return name.normalize("NFD").replace(/\p{M}/gu, "").toLowerCase().split(/[^\p{L}\p{N}]+/u).filter(Boolean);
+}
+function isSameWord(a, b) {
+  if (Math.max(a.length, b.length) > 3)
+    return distance(a, b) <= 1;
+  const [shorter, longer] = a.length <= b.length ? [a, b] : [b, a];
+  return shorter === longer || longer.length === shorter.length + 1 && longer.startsWith(shorter);
+}
+function coversWords(words, pool, lastMayBeCut) {
+  return words.every((word, i) => pool.some((other) => isSameWord(word, other) || lastMayBeCut && i === words.length - 1 && other.startsWith(word)));
+}
+function isWordMatch(query, suggestion) {
+  const matchType = suggestion.match_type ?? suggestion.matchType;
+  if (matchType === "DOMAIN" || matchType === "PERSON")
+    return true;
+  const asked = nameWords(query);
+  const found = nameWords(suggestionName(suggestion));
+  if (asked.length === 0 || found.length === 0)
+    return false;
+  return coversWords(asked, found, true) || coversWords(found, asked, false) || asked.join("") === found.join("");
+}
 function isLeadbayError(error) {
   return typeof error === "object" && error !== null && error.error === true && typeof error.code === "string" && typeof error.message === "string" && typeof error.hint === "string";
 }
@@ -18703,7 +18868,7 @@ async function resolveWithinLens(client, query, lensId) {
 }
 async function resolveAcrossVisibleCorpus(client, query) {
   const suggestions = await client.request("GET", `/search/suggest?q=${encodeURIComponent(query)}`);
-  return suggestions.map((suggestion) => {
+  return suggestions.filter((suggestion) => isWordMatch(query, suggestion)).map((suggestion) => {
     const id = suggestionLeadId(suggestion);
     return {
       id: id ?? "",
@@ -19082,9 +19247,9 @@ var GETTING_STARTED_MANIFEST = {
         leadId: "<the highest-scoring lead id from step 2>"
       },
       forbidden_args: [
-        "enrich \u2014 enrich:true launches a PAID contact reveal off the back of a DRAFT click. They agreed to see an email written, not to spend. Gate 4 is where the reveal gets asked for, explicitly and on its own terms."
+        "enrich \u2014 enrich:true launches a contact reveal, which uses quota, off the back of a DRAFT click. They agreed to see an email written, not to reveal anyone. Gate 4 is where the reveal gets asked for, explicitly and on its own terms."
       ],
-      spend: "This gate spends NOTHING. Call leadbay_prepare_outreach with leadId and nothing else. `recommended_contact` comes back in its post-enrichment shape with email and phone still null \u2014 that is EXPECTED, not a failure, and it is precisely the hook for gate 4: an email written, and nobody to send it to yet. Do not apologise for the missing contact, and do not reach for another tool to fill it in.",
+      spend: "This gate uses NO quota. Call leadbay_prepare_outreach with leadId and nothing else. `recommended_contact` comes back in its post-enrichment shape with email and phone still null \u2014 that is EXPECTED, not a failure, and it is precisely the hook for gate 4: an email written, and nobody to send it to yet. Do not apologise for the missing contact, and do not reach for another tool to fill it in.",
       branches: [
         {
           when: "always",
@@ -19100,7 +19265,7 @@ var GETTING_STARTED_MANIFEST = {
       n: 4,
       gate_label: "Find who to email",
       gate_description: "Reveal the person at that company to send the draft to.",
-      explain: "Point straight at the gap the draft just opened: they have an email ready and nobody to send it to \u2014 it's addressed to a job title, not a person. That's what this step fixes. Explain what enrichment IS: Leadbay can find which roles exist at that company, then reveal the actual human and how to reach them. WHY IT'S USEFUL: they ask for the operations director by name instead of pitching whoever answers the switchboard \u2014 the difference between a conversation and a dead end. Say plainly that the first look is free, and that revealing the contact costs credits and needs their say-so.",
+      explain: "Point straight at the gap the draft just opened: they have an email ready and nobody to send it to \u2014 it's addressed to a job title, not a person. That's what this step fixes. Explain what enrichment IS: Leadbay can find which roles exist at that company, then reveal the actual human and how to reach them. WHY IT'S USEFUL: they ask for the operations director by name instead of pitching whoever answers the switchboard \u2014 the difference between a conversation and a dead end. Say plainly that the first look is free, and that revealing the contact uses a little of their plan's quota and needs their say-so.",
       next_steps: {
         question: "Want to find out who to send that email to?",
         options: [
@@ -19123,8 +19288,8 @@ var GETTING_STARTED_MANIFEST = {
         leadIds: "[<the ONE lead you drafted for at step 3>] \u2014 an ARRAY, always",
         lensId: "<the pinned lens id from step 2>"
       },
-      spend: "TWO BEATS \u2014 free preview FIRST, the real reveal only after the user confirms. Beat 1: call leadbay_enrich_titles with the drafted lead's id + lensId and NO titles / NO confirm / NO email / NO phone. That returns mode:'discover' \u2014 the FREE list of job titles at that company. Say plainly that nothing has been spent yet. Beat 2: name the title the draft is addressed to, tell them BEFORE they decide what it costs (one credit per contact revealed \u2014 here that is ONE contact, one credit), and ask them to confirm. Only then call leadbay_enrich_titles AGAIN with leadIds: [<that lead id>] \u2014 ALWAYS the array, even for a single lead: `leadId` singular is not a key this tool reads, so it is dropped and the paid call falls back to the default wishlist selection, charging for the whole batch \u2014 plus the chosen title, confirm:true and email:true. Poll leadbay_bulk_enrich_status with the returned notification_id and lead_ids until all_done (or the count plateaus), and report the contact that actually resolved. NEVER launch the reveal without an explicit confirm: silence is not consent, and neither is 'they clicked the gate'. If they decline, keep the draft and the title and move on \u2014 that is a normal outcome, not a failure.",
-      quota_note: "After the reveal, close the loop on gate 1 in one line: one credit per contact revealed, so this cost one. Then say the thing that makes it land \u2014 the draft from gate 3 now has a real person and a real address to go to. Re-check leadbay_account_status if you want to show the moved windows. This is where gate 1's numbers stop being abstract: they just watched them move, and got something for it. Keep it to a line; no pricing pitch."
+      spend: "TWO BEATS \u2014 free preview FIRST, the real reveal only after the user confirms. Beat 1: call leadbay_enrich_titles with the drafted lead's id + lensId and NO titles / NO confirm / NO email / NO phone. That returns mode:'discover' \u2014 the FREE list of job titles at that company. Say plainly that nothing has run yet. Beat 2: name the title the draft is addressed to, tell them BEFORE they decide that revealing it uses their plan's quota (here ONE contact; no amount, no price), and ask them to confirm. Only then call leadbay_enrich_titles AGAIN with leadIds: [<that lead id>] \u2014 ALWAYS the array, even for a single lead: `leadId` singular is not a key this tool reads, so it is dropped and the reveal falls back to the default wishlist selection, revealing the whole batch \u2014 plus the chosen title, confirm:true and email:true. Poll leadbay_bulk_enrich_status with the returned notification_id and lead_ids until all_done (or the count plateaus), and report the contact that actually resolved. NEVER launch the reveal without an explicit confirm: silence is not consent, and neither is 'they clicked the gate'. If they decline, keep the draft and the title and move on \u2014 that is a normal outcome, not a failure.",
+      quota_note: "After the reveal, close the loop on gate 1 in one line: the reveal used a little of their plan's quota, never a price. Then say the thing that makes it land \u2014 the draft from gate 3 now has a real person and a real address to go to. Re-check leadbay_account_status if you want to show the moved windows. This is where gate 1's numbers stop being abstract: they just watched them move, and got something for it. Keep it to a line; no pricing pitch."
     }
   ],
   keep_going: KEEP_GOING,
@@ -19198,12 +19363,12 @@ var setQualificationQuestions = {
     properties: {
       questions: {
         type: "array",
-        items: { type: "string" },
+        items: { type: "string", maxLength: 255 },
         description: "Full replacement list of qualification questions (replaces ALL current questions). Mutually exclusive with add/remove."
       },
       add: {
         type: "array",
-        items: { type: "string" },
+        items: { type: "string", maxLength: 255 },
         description: "Questions to append to the current list (deduped). Mutually exclusive with `questions`."
       },
       remove: {
@@ -20058,7 +20223,7 @@ var accountStatus = {
       },
       quota: {
         type: ["object", "null"],
-        description: "Quota state, shaped like the web app. Prefer the `user` group (present for every caller; `org` is admin-only). Each group has `spend[]` and `resources[]` per daily/weekly/monthly window. `spend[]` is the PERCENTAGE-USED + DOLLAR gauge: `current_units`/`max_units` are dollar_cents, so % used = current/max and $ = /100 \u2014 render this like the frontend, never raw 'credits'. `spend[]` is empty for internal/free orgs with no OVERALL_SPEND quota; fall back to the `resources[]` breakdown then. `resources[]` gives per-resource usage: {resource_type, count (used), max_units (per-resource cap or null), window_type, resets_at}. `topup` (when present) is {remaining_cents, total_credit_cents}. Null if /quota_status failed (see quota_error) or genuinely returned nothing. NOTE: the `user`-group preference is for the DISPLAY gauge only \u2014 to pre-check LENS_EXTRA_REFILL before leadbay_extend_lens, look in `quota.org.resources[]` first (admins) and fall back to `quota.user.resources[]` when `org` is absent (non-admins), matching the resource type case-insensitively (LENS_EXTRA_REFILL / lens_extra_refill)."
+        description: "Quota state, shaped like the web app. Prefer the `user` group (present for every caller; `org` is admin-only). Each group has `spend[]` and `resources[]` per daily/weekly/monthly window. `spend[]` is the PERCENTAGE-USED + DOLLAR gauge: `current_units`/`max_units` are dollar_cents, so % used = current/max and $ = /100 \u2014 render this like the frontend, never raw 'credits', and only when the user asks about quota or a window is exhausted. `spend[]` is empty for internal/free orgs with no OVERALL_SPEND quota; fall back to the `resources[]` breakdown then. `resources[]` gives per-resource usage: {resource_type, count (used), max_units (per-resource cap or null), window_type, resets_at}. `topup` (when present) is {remaining_cents, total_credit_cents}. Null if /quota_status failed (see quota_error) or genuinely returned nothing. NOTE: the `user`-group preference is for the DISPLAY gauge only \u2014 to pre-check LENS_EXTRA_REFILL before leadbay_extend_lens, look in `quota.org.resources[]` first (admins) and fall back to `quota.user.resources[]` when `org` is absent (non-admins), matching the resource type case-insensitively (LENS_EXTRA_REFILL / lens_extra_refill)."
       },
       quota_error: {
         type: ["object", "null"],
@@ -22010,7 +22175,7 @@ async function launchOnSelection(client, args, ctx) {
       preview,
       lead_ids: leadIds,
       titles,
-      message: `An identical enrichment was started ${claim.seconds_since}s ago and has not returned its job id yet. Nothing was launched twice and no quota was spent.`,
+      message: `An identical enrichment was started ${claim.seconds_since}s ago and has not returned its job id yet. Nothing was launched twice and no quota was used.`,
       next_action: "Wait a few seconds and call leadbay_enrich_titles again with the same arguments \u2014 it will hand back the job id once the first call settles. Do not treat this as a running job; there is no id to poll yet."
     };
   }
@@ -22028,8 +22193,8 @@ async function launchOnSelection(client, args, ctx) {
       phone,
       notification_id: already.notification_id,
       launched_at: already.launched_at,
-      message: `An identical enrichment was launched ${already.seconds_since}s ago; this call did NOT spend quota again. Poll the original job rather than relaunching.`,
-      next_action: "Poll leadbay_bulk_enrich_status({notification_id, lead_ids, titles, email, phone, include_contacts: true}) until all_done, or until overall_progress.done holds steady across spaced polls (~15-30s apart) \u2014 unresolvable contacts never flip. Pass titles/email/phone every time: they scope counting to the roles and channel THIS run asked for, so a contact enriched earlier cannot report the run as finished. Then report what landed and name what didn't."
+      message: `An identical enrichment was launched ${already.seconds_since}s ago; this call did NOT use quota again. Poll the original job rather than relaunching.`,
+      next_action: "Poll leadbay_bulk_enrich_status({notification_id, lead_ids, titles, email, phone, include_contacts: true}) until all_done, or until overall_progress.done holds steady across spaced polls (~15-30s apart) \u2014 a reservation the provider never answers never flips. Pass titles/email/phone every time: they scope counting to the roles and channel THIS run asked for, so a contact enriched earlier cannot report the run as finished. Then report what landed and name what didn't."
     };
   }
   ctx?.progress?.({
@@ -22066,8 +22231,8 @@ async function launchOnSelection(client, args, ctx) {
     lead_ids: leadIds,
     notification_id: notificationId,
     launched_at: remembered.launched_at,
-    message: notificationId ? "Enrichment job launched (runs async). Unless the user asked NOT to wait, do NOT end your turn here \u2014 poll leadbay_bulk_enrich_status({notification_id, lead_ids, titles, email, phone}) until all_done OR until progress plateaus (overall_progress.done stops climbing across spaced polls \u2014 unresolvable contacts keep all_done:false forever), then report the finished contacts yourself. The notification_id keeps working across conversations and days; completion also surfaces via _meta.notifications / leadbay_account_status.notifications." : "Enrichment job launched, but the backend returned no notification_id, so there is no job id to poll. Poll leadbay_bulk_enrich_status({lead_ids, titles, email, phone}) instead \u2014 it answers per lead without a job id.",
-    next_action: notificationId ? "Unless the user explicitly asked NOT to wait, poll leadbay_bulk_enrich_status({notification_id, lead_ids, titles, email, phone, include_contacts: true}) until all_done \u2014 OR until overall_progress.done holds steady across several SPACED polls (~15-30s apart, ~90s-2min elapsed; don't call a plateau from the first back-to-back reads, and not while partial_failures is present \u2014 that's transient, respect retry_after). Carry titles/email/phone on every poll \u2014 they scope counting to the roles and channel THIS run asked for, and without them a contact enriched months ago counts as done and all_done flips true before anything landed. Then report the resolved enrichment in THIS turn, naming what landed and what didn't. If the user DID ask not to wait, hand back the notification_id \u2014 it resolves later from any conversation." : "Poll leadbay_bulk_enrich_status({lead_ids, titles, email, phone, include_contacts: true}) every ~30s. It scopes counting to these titles and treats a contact as done only when the REQUESTED channel landed, so you do not have to do that yourself. Stop once overall_progress.done stops growing across a couple of spaced re-checks (~90s-2min) \u2014 unresolvable contacts never flip."
+    message: notificationId ? "Enrichment job launched (runs async). Unless the user asked NOT to wait, do NOT end your turn here \u2014 poll leadbay_bulk_enrich_status({notification_id, lead_ids, titles, email, phone}) until all_done OR until progress plateaus (overall_progress.done stops climbing across spaced polls \u2014 a reservation the provider never answers keeps all_done:false forever), then report the finished contacts yourself. The notification_id keeps working across conversations and days; completion also surfaces via _meta.notifications / leadbay_account_status.notifications." : "Enrichment job launched, but the backend returned no notification_id, so there is no job id to poll. Poll leadbay_bulk_enrich_status({lead_ids, titles, email, phone}) instead \u2014 it answers per lead without a job id.",
+    next_action: notificationId ? "Unless the user explicitly asked NOT to wait, poll leadbay_bulk_enrich_status({notification_id, lead_ids, titles, email, phone, include_contacts: true}) until all_done \u2014 OR until overall_progress.done holds steady across several SPACED polls (~15-30s apart, ~90s-2min elapsed; don't call a plateau from the first back-to-back reads, and not while partial_failures is present \u2014 that's transient, respect retry_after). Carry titles/email/phone on every poll \u2014 they scope counting to the roles and channel THIS run asked for, and without them a contact enriched months ago counts as done and all_done flips true before anything landed. Then report the resolved enrichment in THIS turn, naming what landed and what didn't. If the user DID ask not to wait, hand back the notification_id \u2014 it resolves later from any conversation." : "Poll leadbay_bulk_enrich_status({lead_ids, titles, email, phone, include_contacts: true}) every ~30s. It scopes counting to these titles and to the channels this run asked for, and counts a reservation once it has settled, so you do not have to do that yourself. Stop once overall_progress.done stops growing across a couple of spaced re-checks (~90s-2min) \u2014 a reservation the provider never answers never flips."
   };
 }
 async function launchEnrichment(client, args, ctx) {
@@ -22132,14 +22297,14 @@ var enrichTitles = {
       },
       confirm: {
         type: "boolean",
-        description: "Explicit spend decision for the paid enrichment. true = go ahead and launch. false = do NOT spend (a veto: returns mode:'needs_confirmation' and launches nothing, even on hosts without elicitation, and even if an email/phone channel was set). Omitted (and no explicit email/phone channel) \u2192 an elicitation-capable host asks the user before launching; a decline returns mode:'needs_confirmation'. Passing email:true/phone:true also counts as consent."
+        description: "Explicit go-ahead for the email/phone reveal (uses quota). true = go ahead and launch. false = do NOT launch (a veto: returns mode:'needs_confirmation' and launches nothing, even on hosts without elicitation, and even if an email/phone channel was set). Omitted (and no explicit email/phone channel) \u2192 an elicitation-capable host asks the user before launching; a decline returns mode:'needs_confirmation'. Passing email:true/phone:true also counts as consent."
       }
     },
     additionalProperties: false
   },
   outputSchema: {
     type: "object",
-    description: "Branchy return shape; the `mode` (or `status`) field tells the agent which branch it got. Modes: 'discover' (no titles passed), 'preview_only' (no enrichable contacts), 'dry_run', 'needs_confirmation' (paid launch withheld pending user consent), 'already_launched' (idempotent reuse), 'launch_in_flight' (an identical launch is mid-flight and has no id yet), 'launched' (happy path). Status: 'quota_exceeded' (429).",
+    description: "Branchy return shape; the `mode` (or `status`) field tells the agent which branch it got. Modes: 'discover' (no titles passed), 'preview_only' (no enrichable contacts), 'dry_run', 'needs_confirmation' (launch withheld pending user consent), 'already_launched' (idempotent reuse), 'launch_in_flight' (an identical launch is mid-flight and has no id yet), 'launched' (happy path). Status: 'quota_exceeded' (429).",
     properties: {
       mode: {
         type: "string",
@@ -22175,7 +22340,7 @@ var enrichTitles = {
       },
       credits_remaining: {
         type: ["number", "string", "null"],
-        description: `Advisory-only balance (billing.ai_credits), present in discover / preview_only / dry_run modes. Enrichment is gated by QUOTA (email + phone reveals consume the per-window allowance in leadbay_account_status), NOT by this number \u2014 do NOT present it as a spend gate, do NOT say 'you have N credits', and NEVER refuse enrichment because it's low or zero. Confirm the run by naming enrichable_contacts + the channels, not a credit figure. Null = billing unavailable. The string "unlimited" = an internal/unlimited account: proceed freely and say nothing about credits or quota.`
+        description: `Advisory-only balance (billing.ai_credits), present in discover / preview_only / dry_run modes. Enrichment is gated by QUOTA (email + phone reveals consume the per-window allowance in leadbay_account_status), NOT by this number \u2014 do NOT present it as a gate, do NOT say 'you have N credits', and NEVER refuse enrichment because it's low or zero. Confirm the run by naming enrichable_contacts + the channels, not a credit figure. Null = billing unavailable. The string "unlimited" = an internal/unlimited account: proceed freely and say nothing about credits or quota.`
       },
       selected_lead_count: {
         type: "number",
@@ -22369,8 +22534,8 @@ var enrichTitles = {
                 enrichable_contacts: preview2.enrichable_contacts,
                 credits_remaining: await readCreditsRemaining(client),
                 available_titles: availableTitles2,
-                message: "Enrichment not launched \u2014 confirm:false was passed (spend declined). Title & LinkedIn are already on the contact (free); enrichment is the PAID email/phone reveal. Re-call with confirm:true (or email:true) to spend.",
-                next_action: "Re-call leadbay_enrich_titles with confirm:true once the user approves the spend."
+                message: "Enrichment not launched \u2014 confirm:false was passed (declined). Title & LinkedIn are already on the contact (free); enrichment is the email/phone reveal, which uses quota. Re-call with confirm:true (or email:true) to launch.",
+                next_action: "Re-call leadbay_enrich_titles with confirm:true once the user approves the reveal."
               }
             };
           } else if (!willElicit) {
@@ -22449,8 +22614,8 @@ var enrichTitles = {
         enrichable_contacts: preview.enrichable_contacts,
         credits_remaining: creditsRemaining,
         available_titles: availableTitles,
-        message: "Enrichment not launched \u2014 awaiting confirmation. Title & LinkedIn are already on the contact (free); enrichment is the PAID email/phone reveal. Re-call with confirm:true (or email:true) to spend.",
-        next_action: "Confirm the spend with the user, then call leadbay_enrich_titles again with confirm:true."
+        message: "Enrichment not launched \u2014 awaiting confirmation. Title & LinkedIn are already on the contact (free); enrichment is the email/phone reveal, which uses quota. Re-call with confirm:true (or email:true) to launch.",
+        next_action: "Confirm the reveal with the user, then call leadbay_enrich_titles again with confirm:true."
       };
     }
     return await launchEnrichment(client, {
@@ -22510,11 +22675,11 @@ var bulkEnrichStatus = {
       },
       email: {
         type: "boolean",
-        description: "The `email` flag the launch returned. A contact counts as done only once the requested channel has landed."
+        description: "The `email` flag the launch returned. Scopes counting to reservations that asked for email (`enrichment.email_requested`), so a contact reserved by an earlier phone-only run is not counted in this one."
       },
       phone: {
         type: "boolean",
-        description: "The `phone` flag the launch returned. Same rule as `email`."
+        description: "The `phone` flag the launch returned. Same rule as `email`, on `enrichment.phone_requested`."
       },
       include_contacts: {
         type: "boolean",
@@ -22548,7 +22713,7 @@ var bulkEnrichStatus = {
       },
       credits_remaining: {
         type: ["number", "string", "null"],
-        description: `Advisory internal context only \u2014 do NOT display it. It's billing.ai_credits (a CONSUMED counter, not a remaining balance), so it can read 0 on a fresh/quota-backed account and printing _(N credits remaining)_ would falsely say 'out of credits'. Enrichment is gated by QUOTA, not this number. When all_done, show the user's standing by calling leadbay_account_status and rendering its per-window quota gauge \u2014 never a credits line here. ("unlimited" = internal/unlimited account; still nothing to display.) A per-run 'credits used' figure is intentionally NOT returned \u2014 getContacts can't scope cost to this bulk.`
+        description: `Advisory internal context only \u2014 do NOT display it. It's billing.ai_credits (a CONSUMED counter, not a remaining balance), so it can read 0 on a fresh/quota-backed account and printing _(N credits remaining)_ would falsely say 'out of credits'. Enrichment is gated by QUOTA, not this number \u2014 never a credits line here. ("unlimited" = internal/unlimited account; still nothing to display.) A per-run 'credits used' figure is intentionally NOT returned \u2014 getContacts can't scope cost to this bulk.`
       },
       partial_failures: {
         type: "array",
@@ -22599,22 +22764,21 @@ var bulkEnrichStatus = {
     }
     if (leadIds.length > 0) {
       const wantTitles = new Set((params.titles ?? []).map((t) => t.trim().toLowerCase()));
-      const channelResolved = (c) => {
-        if (c?.enrichment?.done !== true)
-          return false;
-        if (params.email && !c.email)
+      const inThisRun = (c) => {
+        if (params.email && c.enrichment.email_requested === false)
           return false;
-        if (params.phone && !c.phone_number)
+        if (params.phone && c.enrichment.phone_requested === false)
           return false;
         return true;
       };
+      const settled = (c) => c.enrichment.done === true;
       let doneSoFar = 0;
       const totalLeads = leadIds.length;
       const results = await pMap(leadIds, async (leadId) => {
         try {
           const out = await getContacts.execute(client, { leadId });
           const contacts = Array.isArray(out?.contacts) ? out.contacts : [];
-          const enrichable = contacts.filter((c) => c && c.enrichment && (wantTitles.size === 0 || typeof c.job_title === "string" && wantTitles.has(c.job_title.trim().toLowerCase())));
+          const enrichable = contacts.filter((c) => c && c.enrichment && inThisRun(c) && (wantTitles.size === 0 || typeof c.job_title === "string" && wantTitles.has(c.job_title.trim().toLowerCase())));
           const fe = Array.isArray(out?._fetch_errors) ? out._fetch_errors : [];
           doneSoFar += 1;
           ctx?.progress?.({
@@ -22633,7 +22797,7 @@ var bulkEnrichStatus = {
           return {
             kind: "ok",
             lead_id: leadId,
-            done: enrichable.filter(channelResolved).length,
+            done: enrichable.filter(settled).length,
             total: enrichable.length,
             ...includeContacts ? { contacts } : {}
           };
@@ -23619,7 +23783,7 @@ async function waitForJob(client, jobId, waitSeconds, ctx, itemsRequested, since
     ctx?.progress?.({
       progress: f.delivered ?? 0,
       total: itemsRequested,
-      message: `${snap.job.state}: ${f.examined ?? 0} examined, ${f.delivered ?? 0} delivered, ${snap.cost.spent}c spent`
+      message: `${snap.job.state}: ${f.examined ?? 0} examined, ${f.delivered ?? 0} delivered`
     });
   }
   return snap;
@@ -23762,7 +23926,7 @@ function mockedSubmitPreview(submit, tool, region) {
       error: true,
       code: "MALFORMED_SUBMIT_RESPONSE",
       message: `${tool}: the submit succeeded but the response carried no job_id, so the job cannot be polled.`,
-      hint: "The job may still be running server-side. Do not re-submit blindly \u2014 reuse the same request_id so a retry dedupes instead of double-spending."
+      hint: "The job may still be running server-side. Do not re-submit blindly \u2014 reuse the same request_id so a retry dedupes instead of launching twice."
     };
   }
   return {
@@ -23774,15 +23938,94 @@ function mockedSubmitPreview(submit, tool, region) {
     region
   };
 }
-function splitItems(snapshot) {
-  const leads = [];
-  const skipped = [];
-  for (const item of snapshot.items) {
-    if (item.status === "skipped")
-      skipped.push(item);
-    else
-      leads.push(item);
+var LEAD_REASON_CHARS = 180;
+var SIGNAL_CHARS = 140;
+var MAX_SIGNALS = 2;
+var MAX_ALTERNATIVE_CONTACTS = 1;
+var RESULT_CHARS = 52e3;
+var LEAN_ROW_CHARS = 700;
+function clip(s, n) {
+  if (typeof s !== "string" || s.length === 0)
+    return void 0;
+  return s.length <= n ? s : s.slice(0, n - 1).trimEnd() + "\u2026";
+}
+function compactContact(c) {
+  if (!c || typeof c !== "object")
+    return c ?? null;
+  const channels = {};
+  for (const [name, state] of Object.entries(c.channels ?? {})) {
+    const s = state;
+    channels[name] = s?.value != null ? { status: s.status, value: s.value } : { status: s?.status };
   }
+  return {
+    lead_contact_id: c.lead_contact_id,
+    name: c.name,
+    role: clip(c.role, 80),
+    linkedin: c.linkedin,
+    channels
+  };
+}
+function compactLead(lead, lean = false) {
+  const company = lead.company ?? {};
+  const fit = lead.fit ?? {};
+  const q = fit.components?.qualification ?? {};
+  const answers = Array.isArray(q.questions) ? q.questions : [];
+  const best = answers.filter((a) => a?.verdict === "yes").sort((a, b) => (b?.score ?? 0) - (a?.score ?? 0))[0];
+  const signals = (Array.isArray(lead.signals) ? lead.signals : []).filter((s) => s?.summary).sort((a, b) => Number(!!b.hot) - Number(!!a.hot) || String(b.date ?? "").localeCompare(String(a.date ?? ""))).slice(0, MAX_SIGNALS).map((s) => ({ summary: clip(s.summary, SIGNAL_CHARS), date: s.date, source_url: s.source_url }));
+  const location = company.location ? { city: company.location.city, region: company.location.region, country: company.location.country } : void 0;
+  return {
+    lead_id: lead.lead_id,
+    company: {
+      name: company.name,
+      website: company.website,
+      location,
+      employees: company.employees,
+      // Kept on trimmed rows too: it is the rendering rules' last fallback for
+      // "why it fits" once the reason and the tags are gone.
+      description: clip(company.description ?? company.short_description, lean ? 80 : 140)
+    },
+    fit: {
+      available: fit.available,
+      score: fit.score,
+      reasoning: lean ? void 0 : clip(best?.reasoning ?? q.ibp?.reasoning, LEAD_REASON_CHARS),
+      components: {
+        qualification: {
+          available: q.available,
+          ai_score: q.ai_score,
+          matched_tags: lean ? void 0 : (Array.isArray(q.matched_tags) ? q.matched_tags : []).map((t) => typeof t === "string" ? t : t?.tag).filter(Boolean)
+        }
+      }
+    },
+    signals: lean ? void 0 : signals,
+    contact: compactContact(lead.contact),
+    alternative_contacts: lean ? void 0 : (Array.isArray(lead.alternative_contacts) ? lead.alternative_contacts : []).slice(0, MAX_ALTERNATIVE_CONTACTS).map(compactContact),
+    // Unbounded, so only full rows carry it: they are measured, trimmed rows
+    // are only reserved for.
+    custom_fields: !lean && Array.isArray(lead.custom_fields) && lead.custom_fields.length > 0 ? lead.custom_fields : void 0,
+    evidence_trimmed: lean ? true : void 0
+  };
+}
+function splitItems(snapshot) {
+  const leanRow = (item) => ({
+    ref: item.ref,
+    status: item.status,
+    status_reason: item.status_reason,
+    ...item.resolution ? { resolution: item.resolution } : {},
+    seq: item.seq,
+    ...item.lead ? { lead: compactLead(item.lead, true) } : {}
+  });
+  const skipped = snapshot.items.filter((i) => i.status === "skipped").map(leanRow);
+  const rows = snapshot.items.filter((i) => i.status !== "skipped");
+  let room = RESULT_CHARS - JSON.stringify(skipped).length;
+  const leads = rows.map((item, i) => {
+    const { from_cache: _fromCache, ...rest } = item;
+    const full = item.lead ? { ...rest, lead: compactLead(item.lead) } : rest;
+    const fullChars = JSON.stringify(full).length;
+    const reserve = (rows.length - i - 1) * LEAN_ROW_CHARS;
+    const out = fullChars + reserve <= room ? full : leanRow(item);
+    room -= out === full ? fullChars : JSON.stringify(out).length;
+    return out;
+  });
   return { leads, skipped };
 }
 function compactBody(body) {
@@ -23870,7 +24113,7 @@ function rejectMalformedExclusions(ids) {
     error: true,
     code: "INVALID_EXCLUDE_LEAD_ID",
     message: `exclude_lead_ids has ${bad.length} entr${bad.length === 1 ? "y" : "ies"} that is not a lead id: ${bad.join(", ")}.`,
-    hint: "Drop or fix those entries and re-call \u2014 every entry must be a non-blank lead id string. Silently skipping them would run the search without an exclusion you asked for, and could re-deliver and charge for that exact lead."
+    hint: "Drop or fix those entries and re-call \u2014 every entry must be a non-blank lead id string. Silently skipping them would run the search without an exclusion you asked for, and could re-deliver that exact lead and use quota on it again."
   };
 }
 function rejectOversizedExclusions(ids) {
@@ -23926,7 +24169,7 @@ function readSpendFlag(value, field) {
     error: true,
     code: "BAD_INPUT",
     message: `${field} must be a boolean (got ${Array.isArray(value) ? "array" : typeof value}: ${JSON.stringify(value)}).`,
-    hint: `Re-call the tool with ${field}: true or ${field}: false as a JSON boolean, not a string or a number. This flag decides whether the user is charged, so an unrecognised value is refused rather than guessed.`
+    hint: `Re-call the tool with ${field}: true or ${field}: false as a JSON boolean, not a string or a number. This flag decides whether the run uses the org's quota, so an unrecognised value is refused rather than guessed.`
   };
 }
 function rejectCountryLocations(locations, region) {
@@ -24044,15 +24287,18 @@ var findNewLeads = {
       },
       qualify: {
         type: "boolean",
-        description: "Run fresh AI qualification and drop candidates scoring below min_ai_score. PAID: ~94 cost_cents per candidate EXAMINED (survivor or not). Default false (free)."
+        description: "Run fresh AI qualification and drop candidates scoring below min_ai_score. Uses the org's usage quota for every candidate EXAMINED (survivor or not). Default false (free)."
       },
       min_ai_score: {
         type: "number",
+        minimum: -30,
+        maximum: 30,
         description: "Disqualification floor on the [-30,+30] qualification DELTA (not the 0-100 fit score). Default 0. Lower to -30 to keep every evaluated lead with its evidence."
       },
       contact_titles: {
         type: "array",
         items: { type: "string" },
+        maxItems: 10,
         description: "Wanted decision-maker titles (max 10), matched semantically cross-language."
       },
       title_gate: {
@@ -24063,7 +24309,7 @@ var findNewLeads = {
       channels: {
         type: "array",
         items: { type: "string", enum: ["email", "phone"] },
-        description: "Contact channels to PURCHASE (email 25c, phone 250c, billed on success only). Empty = free identity tier."
+        description: "Contact channels to find (email, phone). Uses the org's usage quota only when a value is found. Empty = free identity tier."
       },
       exclude_lead_ids: {
         type: "array",
@@ -24077,7 +24323,7 @@ var findNewLeads = {
       },
       max_cost: {
         type: "number",
-        description: "Spend cap for the whole job in cost_cents. Defaults by plan tier (500/2000/5000). The job stops honestly at the cap (stop_reason max_cost)."
+        description: "Usage cap for the whole job, in internal units. Leave it unset: the default (100000) covers any job. What the requested channels need is kept for them; a cap below that is refused, naming the minimum. Never show it to the user as money."
       },
       exploration_cap: {
         type: "number",
@@ -24085,20 +24331,20 @@ var findNewLeads = {
       },
       request_id: {
         type: "string",
-        description: "REQUIRED idempotency key. Derive it from the ask (e.g. 'gyms-texas-2026-07-28'); REUSE the exact same value when retrying the same ask \u2014 a duplicate returns the SAME job instead of double-spending. Use a NEW value only for a genuinely new ask."
+        description: "REQUIRED idempotency key. Derive it from the ask (e.g. 'gyms-texas-2026-07-28'); REUSE the exact same value when retrying the same ask \u2014 a duplicate returns the SAME job instead of launching twice. Use a NEW value only for a genuinely new ask."
       },
       lang: { type: "string", description: "Output language (default: user's language)." },
       confirm: {
         type: "boolean",
-        description: "Explicit spend decision, required only for a PAID search (qualify:true and/or channels). true = the user approved the quote, go ahead. false = a veto (returns mode:'needs_confirmation', spends nothing). Omitted on a paid call \u2192 the tool withholds the submit and returns a free quote to show the user first. The default FREE search (no qualify, no channels) needs no confirm."
+        description: "Explicit go-ahead, required only for a search that uses quota (qualify:true and/or channels). true = the user approved the quote, go ahead. false = a veto (returns mode:'needs_confirmation', uses nothing). Omitted on such a call \u2192 the tool withholds the submit and returns a free quote to show the user first. The default FREE search (no qualify, no channels) needs no confirm."
       },
       dry_run: {
         type: "boolean",
-        description: "Validate + worst-case cost estimate + quota forecast. No job, no spend. Use before the first PAID run of a session."
+        description: "Validate + worst-case usage estimate + quota forecast. No job, uses nothing. Use before the first quota-using run of a session."
       },
       wait_seconds: {
         type: "number",
-        description: "How long to poll before returning (default 45, max 180, 0 = submit + one snapshot). Free searches usually finish inside the window; paid exploration can take minutes \u2014 the result then carries still_running:true and the job_id to check with leadbay_lead_job_status."
+        description: "How long to poll before returning (default 45, max 180, 0 = submit + one snapshot). Free searches usually finish inside the window; qualified exploration can take minutes \u2014 the result then carries still_running:true and the job_id to check with leadbay_lead_job_status."
       }
     },
     required: ["count", "request_id"],
@@ -24200,13 +24446,13 @@ var findNewLeads = {
         submitted: false,
         vetoed,
         paid_because: [
-          buysQualification ? "qualify: true (~94 cost_cents per candidate EXAMINED)" : null,
+          buysQualification ? "qualify: true (uses quota per candidate examined)" : null,
           buysChannels ? `channels requested: ${params.channels.join(", ")}` : null
         ].filter(Boolean),
         quote: forecast,
         estimated_cost: forecast?.estimated_cost ?? null,
         items_requested: forecast?.items_requested ?? null,
-        hint: vetoed ? "confirm:false vetoed the spend \u2014 nothing was submitted. Re-call with confirm:true to proceed, or drop qualify/channels for a free search." : "Show the user this worst-case quote and get an explicit go-ahead, then re-call with confirm:true. For a free search instead: omit qualify and channels.",
+        hint: vetoed ? "confirm:false vetoed the run \u2014 nothing was submitted. Re-call with confirm:true to proceed, or drop qualify/channels for a free search." : "Tell the user what will run and that it uses their plan's quota (no amounts, no money), get an explicit go-ahead, then re-call with confirm:true. For a free search instead: omit qualify and channels.",
         region: client.region
       };
     }
@@ -24413,7 +24659,7 @@ var qualifyLeads = {
       },
       prior_deliveries: {
         type: "object",
-        description: "Selector expanding the org's past MCP deliveries into refs \u2014 billed leads stay re-readable after result expiry. Combine with lead_refs or use alone.",
+        description: "Selector expanding the org's past MCP deliveries into refs \u2014 delivered leads stay re-readable after result expiry. Combine with lead_refs or use alone.",
         properties: {
           job_id: { type: "string" },
           since: { type: "string", description: "ISO instant lower bound." },
@@ -24423,7 +24669,7 @@ var qualifyLeads = {
       },
       qualify: {
         type: "boolean",
-        description: "Fresh AI qualification (default true; ~94 cost_cents per lead needing fresh research+scoring, cache-free when a fresh dossier exists). Owned disqualified leads come back WITH their negative evidence."
+        description: "Fresh AI qualification (default true; uses the org's usage quota for each lead needing fresh research+scoring, none when a fresh dossier exists). Owned disqualified leads come back WITH their negative evidence."
       },
       contact_titles: {
         type: "array",
@@ -24438,24 +24684,24 @@ var qualifyLeads = {
       channels: {
         type: "array",
         items: { type: "string", enum: ["email", "phone"] },
-        description: "Channels to PURCHASE (email 25c, phone 250c, success-only, already-owned values are free). Empty = free identity tier."
+        description: "Channels to find (email, phone). Uses the org's usage quota only when a value is found; already-owned values are free. Empty = free identity tier."
       },
       max_cost: {
         type: "number",
-        description: "Spend cap in cost_cents (plan-tier default when unset)."
+        description: "Usage cap in internal units; leave it unset (default 100000). What the requested channels need is kept for them; a cap below that is refused, naming the minimum. Never show it as money."
       },
       request_id: {
         type: "string",
-        description: "Recommended idempotency key \u2014 REUSE the same value when retrying the same batch so a retry returns the SAME job instead of re-spending."
+        description: "Recommended idempotency key \u2014 REUSE the same value when retrying the same batch so a retry returns the SAME job instead of running it again."
       },
       lang: { type: "string", description: "Output language (default: user's language)." },
       confirm: {
         type: "boolean",
-        description: "Explicit spend decision for the PAID work (fresh qualification and/or channel purchases). true = the user approved the quote, go ahead. false = a veto (returns mode:'needs_confirmation', spends nothing). Omitted on a paid call \u2192 the tool withholds the submit and returns a free quote to show the user first. A fully FREE call (qualify:false and no channels) needs no confirm."
+        description: "Explicit go-ahead for the work that uses quota (fresh qualification and/or channels). true = the user approved, go ahead. false = a veto (returns mode:'needs_confirmation', uses nothing). Omitted on such a call \u2192 the tool withholds the submit and returns a free quote to show the user first. A fully FREE call (qualify:false and no channels) needs no confirm."
       },
       dry_run: {
         type: "boolean",
-        description: "Validate + worst-case cost + quota forecast. No job, no spend."
+        description: "Validate + worst-case usage + quota forecast. No job, uses nothing."
       },
       wait_seconds: {
         type: "number",
@@ -24515,7 +24761,7 @@ var qualifyLeads = {
         quote: forecast,
         estimated_cost: forecast?.estimated_cost ?? null,
         items_requested: forecast?.items_requested ?? null,
-        hint: vetoed ? "confirm:false vetoed the spend \u2014 nothing was submitted. Re-call with confirm:true to proceed, or qualify:false with no channels for a free pass." : "Show the user this worst-case quote and get an explicit go-ahead, then re-call with confirm:true. For a free pass instead: qualify:false and no channels.",
+        hint: vetoed ? "confirm:false vetoed the run \u2014 nothing was submitted. Re-call with confirm:true to proceed, or qualify:false with no channels for a free pass." : "Tell the user what will run and that it uses their plan's quota (no amounts, no money), get an explicit go-ahead, then re-call with confirm:true. For a free pass instead: qualify:false and no channels.",
         region: client.region
       };
     }
@@ -25116,7 +25362,7 @@ var newLens = {
   inputSchema: {
     type: "object",
     properties: {
-      name: { type: "string", description: "Display name for the new lens (required)." },
+      name: { type: "string", maxLength: 255, description: "Display name for the new lens (required)." },
       sectors: {
         type: "array",
         items: { type: "string" },
@@ -25149,7 +25395,7 @@ var newLens = {
         type: "number",
         description: "Lens id to clone from. Defaults to the active/default lens."
       },
-      description: { type: "string", description: "Optional lens description." },
+      description: { type: "string", maxLength: 255, description: "Optional lens description." },
       confirm: {
         type: "boolean",
         description: "Safety gate. Defaults to false \u2192 the tool returns a PREVIEW and creates nothing. Show the preview to the user, get their explicit go-ahead, then re-call the SAME args with confirm:true to actually create the lens."
@@ -26041,8 +26287,8 @@ var sendFeedback = {
 
 // ../core/dist/artifact-runtime.generated.js
 var ARTIFACT_KIT_VERSION = "0.5.0";
-var ARTIFACT_RUNTIME = '"use strict";(()=>{var L=Object.defineProperty;var A=(e,r,t)=>r in e?L(e,r,{enumerable:!0,configurable:!0,writable:!0,value:t}):e[r]=t;var a=(e,r,t)=>A(e,typeof r!="symbol"?r+"":r,t);var _=`\n:root{\n--color-black:#191919;--color-white:#fff;\n--color-gray-1:#f9f9f9;--color-gray-2:#f0f0f0;--color-gray-3:#e0e0e0;--color-gray-4:#cecece;\n--color-gray-5:#c4c4c4;--color-gray-6:#8d8d8d;--color-gray-7:#787878;--color-gray-8:#646464;\n--color-gray-9:#202020;\n--color-linkedin:#0a66c2;\n--color-blue-background:oklch(0.947 0.029 251);--color-blue-foreground:oklch(0.564 0.181 251);\n--color-green-background:oklch(0.947 0.029 141);--color-green-foreground:oklch(0.564 0.181 141);\n--color-red-background:oklch(0.947 0.029 26);--color-red-foreground:oklch(0.564 0.191 26);\n--color-gold-background:oklch(0.972 0.049 91);--color-gold-foreground:oklch(0.667 0.177 91);\n--color-cherry-background:oklch(0.947 0.029 15);--color-cherry-foreground:oklch(0.44 0.146 15);\n--color-red-like:var(--color-cherry-foreground);\n--lb-font:"Nikkei Maru",system-ui,-apple-system,"Segoe UI",Roboto,sans-serif;\n--lb-radius:1rem;--lb-radius-sm:0.625rem;--lb-gap:0.75rem;\n--lb-surface:var(--color-gray-1);--lb-border:var(--color-gray-3);\n--lb-fg:var(--color-black);--lb-muted:var(--color-gray-8);--lb-field:var(--color-white);\n}\n:root[data-theme=dark],:root[data-lb-theme=dark]{\n--lb-surface:var(--color-gray-9);--lb-border:var(--color-gray-8);\n--lb-fg:var(--color-white);--lb-muted:var(--color-gray-3);--lb-field:var(--color-gray-9);\n}\n@media(prefers-color-scheme:dark){:root:not([data-theme=light]):not([data-lb-theme=light]){\n--lb-surface:var(--color-gray-9);--lb-border:var(--color-gray-8);\n--lb-fg:var(--color-white);--lb-muted:var(--color-gray-3);--lb-field:var(--color-gray-9);\n}}\n.lb-card{display:grid;gap:var(--lb-gap);padding:0.875rem;\nbackground-color:var(--lb-surface);border:1px solid var(--lb-border);\nborder-radius:var(--lb-radius);corner-shape:squircle;color:var(--lb-fg);\nfont-family:var(--lb-font);\nbox-shadow:0 1rem 2.5rem color-mix(in srgb,var(--color-gray-9) 6%,transparent),\n0 0.125rem 0.5rem color-mix(in srgb,var(--color-gray-9) 4%,transparent)}\n.lb-card-head{display:flex;justify-content:space-between;align-items:baseline;gap:var(--lb-gap)}\n.lb-title{font-size:0.875rem;font-weight:600;line-height:1.25rem;color:var(--lb-fg)}\n.lb-sub{font-size:0.8125rem;line-height:1.125rem;color:var(--lb-muted)}\n.lb-row{display:flex;align-items:center;gap:var(--lb-gap);flex-wrap:wrap}\n.lb-stack{display:grid;gap:var(--lb-gap)}\n.lb-select,.lb-input{font:inherit;font-family:var(--lb-font);font-size:0.8125rem;color:var(--lb-fg);\nbackground-color:var(--lb-field);border:1px solid var(--lb-border);\nborder-radius:var(--lb-radius-sm);corner-shape:squircle;padding:0.4rem 0.55rem;min-height:2.125rem}\n.lb-btn{font:inherit;font-family:var(--lb-font);font-size:0.8125rem;font-weight:600;\ncolor:var(--lb-fg);background-color:var(--lb-field);border:1px solid var(--lb-border);\nborder-radius:var(--lb-radius-sm);corner-shape:squircle;padding:0.4rem 0.85rem;min-height:2.125rem;\ncursor:pointer;transition:background-color .15s,border-color .15s,color .15s}\n.lb-btn:hover:not([disabled]){border-color:var(--color-gray-6)}\n.lb-btn:focus-visible,.lb-select:focus-visible,.lb-input:focus-visible{\noutline:2px solid var(--color-blue-foreground);outline-offset:1px}\n.lb-btn[data-lb-state=loading]{opacity:.55;cursor:progress}\n.lb-btn[data-lb-state=success]{background-color:var(--color-green-background);\nborder-color:var(--color-green-foreground);color:var(--color-green-foreground)}\n.lb-btn[data-lb-state=error],.lb-select[data-lb-state=error]{\nbackground-color:var(--color-red-background);border-color:var(--color-red-foreground);\ncolor:var(--color-red-foreground)}\n.lb-btn[data-lb-state=unavailable],.lb-btn[disabled]{opacity:.5;cursor:not-allowed}\n.lb-msg{font-size:0.8125rem;line-height:1.125rem;color:var(--lb-muted)}\n.lb-msg[data-tone=error]{color:var(--color-red-foreground)}\n.lb-msg[data-tone=ok]{color:var(--color-green-foreground)}\n.lb-chip{display:inline-flex;align-items:center;gap:.25rem;white-space:nowrap;\nfont-size:0.75rem;font-weight:600;line-height:1rem;padding:0.125rem 0.5rem;\nborder-radius:var(--lb-radius-sm);corner-shape:squircle;\nbackground-color:var(--color-gray-2);color:var(--lb-muted)}\n.lb-chip[data-status=WANTED]{background-color:var(--color-blue-background);color:var(--color-blue-foreground)}\n.lb-chip[data-status=WON]{background-color:var(--color-green-background);color:var(--color-green-foreground)}\n.lb-chip[data-status=LOST]{background-color:var(--color-red-background);color:var(--color-red-foreground)}\n.lb-chip[data-status=UNWANTED]{background-color:var(--color-gray-2);color:var(--color-gray-7)}\n.lb-chip[data-taste=liked]{background-color:var(--color-cherry-background);color:var(--color-red-like)}\n.lb-chip[data-taste=disliked]{background-color:var(--color-gray-2);color:var(--color-gray-7)}\n.lb-chips{display:flex;align-items:center;gap:.35rem;flex-wrap:wrap}\n.lb-chip[hidden]{display:none}\n.lb-table{width:100%;border-collapse:collapse;font-family:var(--lb-font);color:var(--lb-fg)}\n.lb-table th,.lb-table td{text-align:left;padding:0.5rem 0.4rem;\nborder-bottom:1px solid var(--lb-border);vertical-align:middle;font-size:0.8125rem}\n.lb-table th{font-size:0.75rem;font-weight:600;color:var(--lb-muted);\ntext-transform:uppercase;letter-spacing:.04em}\n.lb-link{color:var(--color-blue-foreground);text-decoration:none}\n.lb-link:hover{text-decoration:underline}\n/* Quiet text link out of the artifact. Button-height so it shares the row\'s\n   baseline; understated so it never competes with the actions beside it. The\n   arrow is a bare diagonal stroke \\u2014 an escape-hatch marker, not an icon that\n   asks to be read. */\n.lb-link-out{display:inline-flex;align-items:center;gap:.3rem;\nfont-size:0.75rem;line-height:1rem;min-height:2.125rem;\ncolor:var(--lb-fg);text-decoration:none;opacity:.65;transition:opacity .15s}\n.lb-link-out:hover{opacity:1;text-decoration:underline}\n.lb-link-out:focus-visible{outline:2px solid var(--color-blue-foreground);outline-offset:1px;\nborder-radius:var(--lb-radius-sm)}\n.lb-link-out svg{width:.85em;height:.85em;flex-shrink:0}\n/* Pushes whatever follows it to the right edge of an .lb-row, so a trailing\n   link sits on the SAME baseline as the row\'s buttons instead of on its own\n   line below them. */\n.lb-spacer{flex:1 1 auto}\n.lb-spinner{display:inline-block;width:.7em;height:.7em;border:2px solid var(--lb-border);\nborder-top-color:var(--color-blue-foreground);border-radius:50%;animation:lb-spin .8s linear infinite}\n@keyframes lb-spin{to{transform:rotate(1turn)}}\n@media(prefers-reduced-motion:reduce){.lb-spinner{animation:none}\n.lb-btn{transition-property:none}}\n`,p="lb-styles";var x="0.5.0",c=class extends Error{constructor(t,o={}){super(t);a(this,"code");a(this,"raw");this.name="LbError",this.code=o.code,this.raw=o.raw}},y=null,f=3e4;function C(){let e=globalThis.cowork;return e&&typeof e.callMcpTool=="function"?(r,t)=>e.callMcpTool(r,t):null}function S(e){if(e&&typeof e=="object"&&"content"in e){let r=e.content;if(Array.isArray(r)&&r[0]&&typeof r[0].text=="string")return r[0].text}return null}function E(e){if(!e||typeof e!="object")return e;let r=e;if(r.isError)throw new c(S(e)??"tool call failed",{raw:e});if("structuredContent"in r&&r.structuredContent!=null)return r.structuredContent;let t=S(e);if(t!=null)try{return JSON.parse(t)}catch{return t}return e}function I(e){return e instanceof Error?e.message:String(e)}function v(e){let r=e instanceof c?e.code:void 0;return{message:I(e),unavailable:r==="unavailable",code:r}}function O(e={}){y=e.call??null,f=e.timeoutMs??3e4}function R(){if(typeof document>"u"||!document.head)return null;let e=document.getElementById(p);if(e)return e;let r=document.createElement("style");return r.id=p,r.textContent=_,document.head.appendChild(r),r}async function T(e,r){if(!f||f<=0)return e;let t,o=new Promise((n,i)=>{t=setTimeout(()=>i(new c(`"${r}" timed out after ${f}ms`,{code:"timeout"})),f)});try{return await Promise.race([e,o])}finally{t&&clearTimeout(t)}}async function l(e,r={}){if(y)return E(await T(Promise.resolve(y(e,r)),e));let t=C();if(!t)throw new c("Leadbay bridge unavailable (window.cowork absent)",{code:"unavailable"});return E(await T(Promise.resolve(t(e,r)),e))}var g=class{constructor(){a(this,"subs",new Set)}subscribe(r){return this.subs.add(r),r(this),()=>this.subs.delete(r)}emit(){for(let r of this.subs)r(this)}};function N(e){return Array.isArray(e)?e.map(r=>r&&typeof r=="object"?r:{value:r,label:String(r)}):[]}var b=class extends g{constructor(t={}){super();a(this,"kind");a(this,"value");a(this,"options",[]);a(this,"loading",!1);a(this,"error",null);a(this,"ready",!1);a(this,"cfg");a(this,"depUnsubs",[]);a(this,"seq",0);this.cfg=t,this.kind=t.kind,this.value=t.value??"";for(let o of t.dependsOn??[]){let n=o.value;this.depUnsubs.push(o.subscribe(()=>{o.value!==n&&(n=o.value,this.cfg.load&&this.load())}))}t.load&&(t.autoLoad??!0)&&this.load()}async load(){if(!this.cfg.load)return;let t=++this.seq;this.loading=!0,this.error=null,this.emit();try{let o=await this.cfg.load();if(t!==this.seq)return;this.options=this.cfg.options?this.cfg.options(o):N(o),this.ready=!0;let n=this.value==null?"":String(this.value);this.options.length&&(n===""||!this.options.some(i=>String(i.value)===n))&&(this.value=this.options[0].value)}catch(o){if(t!==this.seq)return;this.options=[],this.error=v(o)}finally{t===this.seq&&(this.loading=!1,this.emit())}}setValue(t){this.value=t;let o=this.validate();this.error=o?{message:o,unavailable:!1}:null,this.emit()}validate(){return this.cfg.validate?this.cfg.validate(this.value):null}get valid(){return this.validate()==null}reset(){this.value=this.cfg.value??"",this.error=null,this.emit()}dispose(){for(let t of this.depUnsubs)t();this.depUnsubs=[]}};function P(e){if(!e||typeof e!="object")return null;let r=e;if(r.error!==!0)return null;let t=typeof r.message=="string"&&r.message?r.message:"tool call failed",o=typeof r.hint=="string"&&r.hint?` \\u2014 ${r.hint}`:"";return`${t}${o}`}var s=class extends g{constructor(t){super();a(this,"loading",!1);a(this,"error",null);a(this,"lastResult",null);a(this,"cfg");this.cfg=t}async run(){if(this.loading)return;for(let n of this.cfg.fields??[]){let i=n.validate();if(i!=null){this.error={message:i,unavailable:!1},this.emit();return}}if(this.cfg.confirm&&typeof globalThis.confirm=="function"&&!globalThis.confirm(this.cfg.confirm))return;this.loading=!0,this.error=null,this.emit();let t;try{let n=typeof this.cfg.args=="function"?this.cfg.args():this.cfg.args??{};t=await l(this.cfg.tool,n)}catch(n){this.error=v(n),this.loading=!1,this.emit(),this.cfg.onError?.(this.error);return}let o=P(t)??this.cfg.checkResult?.(t)??null;if(o!=null){this.error={message:o,unavailable:!1},this.loading=!1,this.emit(),this.cfg.onError?.(this.error);return}return this.lastResult=t,this.loading=!1,this.emit(),this.cfg.onSuccess?.(t),t}reset(){this.error=null,this.lastResult=null,this.emit()}},u=class extends g{constructor(t){super();a(this,"data",null);a(this,"loading",!1);a(this,"refreshing",!1);a(this,"error",null);a(this,"done",!1);a(this,"cfg");a(this,"timer",null);a(this,"seq",0);this.cfg=t,(t.autoLoad??!0)&&this.load()}async load(){this.clearTimer();let t=++this.seq;this.data==null?this.loading=!0:this.refreshing=!0,this.error=null,this.emit();try{let n=await this.cfg.load();if(t!==this.seq)return;this.data=n,this.done=this.cfg.until?this.cfg.until(n):!0,this.cfg.pollEvery&&!this.done&&(this.timer=setTimeout(()=>void this.load(),this.cfg.pollEvery))}catch(n){if(t!==this.seq)return;this.error=v(n)}finally{t===this.seq&&(this.loading=!1,this.refreshing=!1,this.emit())}}refresh(){return this.load()}stop(){this.clearTimer()}clearTimer(){this.timer&&(clearTimeout(this.timer),this.timer=null)}},h=class extends g{constructor(t){super();a(this,"items",[]);a(this,"page",0);a(this,"pageSize");a(this,"total",0);a(this,"loading",!1);a(this,"error",null);a(this,"cfg");a(this,"seq",0);this.cfg=t,this.pageSize=t.pageSize??20,(t.autoLoad??!0)&&this.loadPage(0)}async loadPage(t){let o=++this.seq;this.loading=!0,this.error=null,this.emit();try{let n=await this.cfg.load({page:t,pageSize:this.pageSize});if(o!==this.seq)return;this.items=n.items??[],this.total=n.total??this.items.length,this.page=t}catch(n){if(o!==this.seq)return;this.error=v(n)}finally{o===this.seq&&(this.loading=!1,this.emit())}}next(){return this.loadPage(this.page+1)}prev(){return this.loadPage(Math.max(0,this.page-1))}get hasMore(){return(this.page+1)*this.pageSize<this.total}};function M(e,r){let t=r.error?.unavailable?"unavailable":r.loading?"loading":r.error?"error":"ready";e.setAttribute("data-lb-state",t),r.error?e.setAttribute("data-lb-error",r.error.message):e.removeAttribute("data-lb-error")}function D(e,r){let t=()=>r.setValue(e.value);e.addEventListener("change",t);let o=r.subscribe(()=>{M(e,r),e.disabled=r.loading,e.innerHTML="";for(let n of r.options){let i=document.createElement("option");i.value=String(n.value),i.textContent=n.label,e.appendChild(i)}e.value=r.value==null?"":String(r.value)});return()=>{e.removeEventListener("change",t),o()}}function z(e,r){let t=e.type==="checkbox",o=e.tagName==="SELECT"?"change":"input",n=()=>r.setValue(t?e.checked:e.value);e.addEventListener(o,n);let i=r.subscribe(()=>{if(t)e.checked=!!r.value;else{let d=r.value==null?"":String(r.value);e.value!==d&&(e.value=d)}e.setAttribute("data-lb-state",r.error?"error":"ready"),r.error?e.setAttribute("data-lb-error",r.error.message):e.removeAttribute("data-lb-error")});return()=>{e.removeEventListener(o,n),i()}}function F(e,r){let t=n=>{n.preventDefault(),r.run()};e.addEventListener("click",t);let o=r.subscribe(()=>{let n=r.error?.unavailable?"unavailable":r.loading?"loading":r.error?"error":r.lastResult!=null?"success":"idle";e.setAttribute("data-lb-state",n),"disabled"in e&&(e.disabled=r.loading),r.error?e.setAttribute("data-lb-error",r.error.message):e.removeAttribute("data-lb-error")});return()=>{e.removeEventListener("click",t),o()}}var q=["STILL_CHASING","COULD_NOT_REACH_STILL_TRYING","INTEREST_VALIDATED_OR_MEETING_PLANED","NOT_INTERESTED_LOST"],k=[{value:"",label:"Default ranking"},{value:"SCORE:DESC",label:"Score \\u2193"},{value:"SCORE:ASC",label:"Score \\u2191"},{value:"NAME:ASC",label:"Name A\\u2192Z"},{value:"NAME:DESC",label:"Name Z\\u2192A"},{value:"SIZE:DESC",label:"Size \\u2193"},{value:"SIZE:ASC",label:"Size \\u2191"},{value:"SECTOR:ASC",label:"Sector A\\u2192Z"},{value:"STATUS:ASC",label:"Status A\\u2192Z"},{value:"CONTACT_COUNT:DESC",label:"Contacts \\u2193"},{value:"LAST_PROSPECTING_ACTION_AT:DESC",label:"Last action \\u2193"},{value:"LAST_PROSPECTING_ACTION_AT:ASC",label:"Last action \\u2191"},{value:"EPILOGUE_STATUS_SET_AT:DESC",label:"Outcome set \\u2193"},{value:"LIKED:DESC",label:"Liked first"},{value:"DISLIKED:DESC",label:"Disliked first"}];function U(e){let r=String(e??"").trim().toUpperCase(),t=k.some(o=>o.value===r);return new b({kind:"select",value:t?r:"",load:async()=>k.slice()})}var m=[{value:"WANTED",label:"Wanted"},{value:"WON",label:"Won"},{value:"LOST",label:"Lost"},{value:"UNWANTED",label:"Unwanted"}],H={value:"",label:"\\u2014 Not set \\u2014"};function j(e){let r=String(e??"").trim().toUpperCase(),t=m.some(o=>o.value===r);return new b({kind:"select",value:t?r:"",validate:o=>String(o??"")===""?"Pick a status":null,load:async()=>t?m.slice():[H,...m]})}function W(e){let r=()=>{let t=typeof e.leadIds=="function"?e.leadIds():e.leadIds;return Array.isArray(t)?t:e.leadId?[e.leadId]:[]};return new s({tool:"leadbay_set_lead_status",fields:e.date?[e.status,e.date]:[e.status],confirm:e.confirm,args:()=>({lead_ids:r(),status:e.status.value,...e.date&&e.date.value?{status_date:e.date.value}:{},...e.ask?{_triggered_by:e.ask}:{}}),checkResult:t=>{let o=t?.failed;if(!Array.isArray(o)||o.length===0)return null;let n=r().length,i=o[0]?.message??"write rejected";return o.length===n?`Status not applied: ${i}`:`${o.length} of ${n} leads failed: ${i}`}})}function $(e){return new b({kind:"select",load:()=>l("leadbay_list_campaigns",{_triggered_by:e}),options:r=>(r?.campaigns??[]).map(o=>{let n=o?.campaign??o;return n?.id?{value:n.id,label:n.name??n.ai_generated_name??String(n.id)}:null}).filter(o=>o!=null)})}function V(e){return new s({tool:"leadbay_report_outreach",fields:e.note?[e.note]:[],args:()=>({lead_id:e.leadId,...e.status?{epilogue_status:e.status.value}:{},note:e.note?e.note.value:"",verification:{source:"user_confirmed",ref:e.ref??"logged from artifact"},_triggered_by:e.ask})})}function G(e){return new s({tool:"leadbay_add_note",fields:[e.note],args:()=>({leadId:e.leadId,note:e.note.value})})}function Z(e){return new s({tool:"leadbay_like_lead",args:{lead_id:e}})}function Y(e){return new s({tool:"leadbay_dislike_lead",args:{lead_id:e}})}function B(e,r){return new u({autoLoad:!1,load:()=>l("leadbay_account_history",{leadId:e,_triggered_by:r})})}function K(e,r){return new u({autoLoad:!1,load:()=>l("leadbay_research_lead_by_id",{leadId:e,_triggered_by:r})})}function J(e){let r=null;return new u({...e.autoLoad!==void 0?{autoLoad:e.autoLoad}:{},pollEvery:e.pollEvery??4e3,until:t=>!!t?.all_done,load:async()=>{if(!r){let t=await l("leadbay_enrich_titles",{...e.leadIds?{leadIds:e.leadIds}:{},titles:e.titles,...e.email!==void 0?{email:e.email}:{},...e.phone!==void 0?{phone:e.phone}:{},...e.confirm!==void 0?{confirm:e.confirm}:{},_triggered_by:e.ask}),o=Array.isArray(t?.lead_ids)?t.lead_ids:[],n=t?.notification_id??null;if(r=n||o.length>0?{notification_id:n,lead_ids:o}:null,!r)return{...t,all_done:!0,no_job:!0}}return l("leadbay_bulk_enrich_status",{...r.notification_id?{notification_id:r.notification_id}:{},...r.lead_ids.length>0?{lead_ids:r.lead_ids}:{},...e.titles?{titles:e.titles}:{},...e.email!==void 0?{email:e.email}:{},...e.phone!==void 0?{phone:e.phone}:{},_triggered_by:e.ask})}})}function Q(e){let r=()=>typeof e.order=="string"?e.order:String(e.order?.value??"");return new h({pageSize:e.pageSize??20,load:async({page:t,pageSize:o})=>{let n=await l("leadbay_pull_leads",{page:t,count:o,...e.lensId?{lensId:e.lensId}:{},...r()?{order:r()}:{},_triggered_by:e.ask}),i=n.leads??[];return{items:i,total:n.pagination?.total??i.length}}})}function X(e){let r=e.source??"followups",t=()=>typeof e.order=="string"?e.order:String(e.order?.value??"");return new h({pageSize:e.pageSize??20,load:async({page:o,pageSize:n})=>{let d=r==="campaign"?await l("leadbay_campaign_call_sheet",{campaign_id:e.campaignId,page:o,count:n,_triggered_by:e.ask}):await l("leadbay_pull_followups",{page:o,count:n,...e.city?{city:e.city}:{},...t()?{order:t()}:{},_triggered_by:e.ask}),w=d.leads??d.items??[];return{items:w,total:d.total_leads??d.pagination?.total??w.length}}})}function ee(e){return new u({load:()=>l("leadbay_team_activity",{weeks:e.weeks??4,_triggered_by:e.ask})})}var re={VERSION:x,configure:O,styles:R,call:l,field:e=>new b(e),action:e=>new s(e),resource:e=>new u(e),list:e=>new h(e),bindSelect:D,bindValue:z,bindAction:F,campaigns:$,outreach:V,note:G,like:Z,dislike:Y,leadStatus:j,setStatus:W,sortOrder:U,leadHistory:B,leadProfile:K,enrichment:J,callList:X,leadList:Q,teamActivity:ee,EPILOGUE_STATUSES:q,LEAD_STATUSES:m,SORT_ORDERS:k};typeof globalThis<"u"&&(globalThis.LeadbayArtifacts=re);})();';
-var ARTIFACT_USAGE_GUIDE = '# Leadbay Artifact Kit \u2014 headless domain components\n\nYou are building a single-file HTML **artifact** the user runs inside cowork. This\nkit gives you **headless view-models** that own a control\'s whole data lifecycle \u2014\nload/populate from a Leadbay call, hold value/state, poll, validate, and\nencapsulate the API call + business rules. **You own 100% of markup/layout/style.**\nThe library renders nothing. Inline the runtime once as a `<script>`; it exposes\none global `window.LeadbayArtifacts` (call it `lb`). Vanilla, no React, no build.\n\nPass every tool you use as the artifact\'s `mcp_tools` so the host permits it.\n\n## Two layers\n\n**Primitives** (generic):\n- `lb.field({ load, options, value, validate, dependsOn })` \u2014 a value + optionally\n  API-populated options. `.value/.setValue/.options/.loading/.error/.valid/.subscribe`.\n- `lb.action({ tool, args, fields, confirm, onSuccess, onError })` \u2014 a write/submit.\n  `.run()/.loading/.error/.lastResult/.subscribe`.\n- `lb.resource({ load, pollEvery?, until?, autoLoad? })` \u2014 one read that may change:\n  load-on-click or poll-until-`until`. `.data/.loading/.refreshing/.error/.done/.load()/.refresh()/.stop()/.subscribe`.\n- `lb.list({ load, pageSize })` \u2014 paginated rows. `.items/.page/.total/.loading/.loadPage(n)/.next()/.prev()/.hasMore/.subscribe`.\n\n`.error` is `{ message, unavailable } | null`. `subscribe(cb)` fires immediately\nthen on every change \u2014 render your own DOM from it.\n\n**Domain components** (pre-wired \u2014 bake in the tool name, arg shape, and footguns):\n\n| Call | Returns | For |\n|---|---|---|\n| `lb.campaigns(ask)` | field | a campaign `<select>`, options from `leadbay_list_campaigns` |\n| `lb.outreach({leadId, ask, status?, note?})` | action | log a call \u2192 `report_outreach` (verification + `_triggered_by` baked in) |\n| `lb.note({leadId, note})` | action | add a note \u2192 `add_note` |\n| `lb.like(leadId)` / `lb.dislike(leadId)` | action | taste signal |\n| `lb.leadStatus(current?)` | field | a status `<select>` (Wanted/Won/Lost/Unwanted) |\n| `lb.setStatus({leadId or leadIds, status, date?, ask})` | action | write the org CRM status \u2192 `set_lead_status` |\n| `lb.leadHistory(leadId, ask)` | resource (lazy) | notes + activities + engagement \u2192 `account_history` |\n| `lb.leadProfile(leadId, ask)` | resource (lazy) | full lead profile \u2192 `research_lead_by_id` |\n| `lb.sortOrder(current?)` | field | a sort `<select>` mirroring the app\'s TableSort |\n| `lb.leadList({lensId?, order?, ask})` | list | a sortable Discover batch \u2192 `pull_leads` |\n| `lb.callList({source:\'followups\'\\|\'campaign\', campaignId?, city?, ask})` | list | a cold-call list (Monitor or a campaign) |\n| `lb.enrichment({leadIds, titles, ask, pollEvery?})` | resource (polling) | launch + watch contact enrichment |\n| `lb.teamActivity({weeks, ask})` | resource | manager leaderboard + activity trend \u2192 `leadbay_team_activity` |\n\n`lb.EPILOGUE_STATUSES` = the 4 disposition values\n(`STILL_CHASING`, `COULD_NOT_REACH_STILL_TRYING`, `INTEREST_VALIDATED_OR_MEETING_PLANED`, `NOT_INTERESTED_LOST`).\n`lb.LEAD_STATUSES` = the 4 org CRM statuses as `{value,label}` (`WANTED`, `WON`, `LOST`, `UNWANTED`).\n`lb.SORT_ORDERS` = the sort options as `{value,label}`; values are the backend `FIELD:ASC|DESC` enum.\n\n**Sorting is a SERVER concern.** `lb.leadList` and `lb.callList` take an `order`\n(a `lb.sortOrder()` field or a literal) and send it upstream; the backend sorts\nthe whole lens / Monitor and returns the requested page of that. Never re-sort\nrows in the browser \u2014 you would be reordering one page of a larger set, showing\nleads that do not belong at that position. The empty value means "no order\nparam", i.e. the tab\'s own ranking, which is the right default. Changing the\nsort should reset to page 0. Campaign call sheets cannot sort:\n`leadbay_campaign_call_sheet` has no `order` param, and `lb.callList` drops it\nfor that source rather than sending something the tool would reject.\n\n**Two different systems.** Epilogue = how one outreach attempt went (drives\nfollow-up ranking). Lead status = the commercial outcome, org-wide \u2014 the same\nfield the website\'s status selector writes. A won deal is a LEAD STATUS;\n"she didn\'t pick up" is an EPILOGUE. Setting one never sets the other, so when\nthe user reports both in one breath, fire both actions.\n\n**Binding sugar** (optional; binds a view-model to YOUR native element, no style):\n`lb.bindSelect(selectEl, field)` (populates options + value), `lb.bindValue(inputEl, field)`,\n`lb.bindAction(buttonEl, action)`. They set `data-lb-state`\n(`ready|loading|error|success|unavailable`) + `data-lb-error` on your element as\nstyling hooks. For lists/resources, use `.subscribe()` and render yourself.\n\n`ask` is the user\'s request this artifact serves \u2014 it becomes `_triggered_by`.\n\n## The skin (optional) \u2014 `lb.styles()`\n\nCall it once and you get a small `lb-*` stylesheet, so every artifact you build\nshares one visual language instead of re-inventing padding and colours. It is\n**opt-in**: skip it and you get exactly the unstyled HTML you wrote. It injects\nno markup and never touches your `class` attributes.\n\n```js\nlb.styles();   // idempotent \u2014 safe to call per row\n```\n\n| Class | For |\n|---|---|\n| `lb-card` / `lb-card-head` / `lb-title` / `lb-sub` | a lead card + its header |\n| `lb-row` / `lb-stack` / `lb-spacer` | control row / vertical spacing / flex filler that right-aligns what follows |\n| `lb-link-out` | quiet external link (icon inherits currentColor) \u2014 "Open in Leadbay" |\n| `lb-select` / `lb-input` / `lb-btn` | form controls (state-aware, see below) |\n| `lb-msg` (`data-tone="error\\|ok"`) | inline feedback |\n| `lb-chip` (`data-status="WON\\|LOST"`) | a status pill |\n| `lb-table` | leads table |\n| `lb-spinner` | inline busy indicator |\n\nControls react to the `data-lb-state` the bind helpers already set \u2014 a bound\n`lb-btn` dims while loading, goes green on success, red on error, all with no\nextra CSS from you.\n\nThe palette is the **product design system**, ported from\n`frontend/packages/style/color.css` \u2014 same `--color-gray-1\u20269` ramp, same\nsemantic `--color-{green,red,blue,gold}-{background,foreground}` pairs, same\n`1rem` / `0.625rem` radii and `corner-shape: squircle` as the app\'s components.\nAn artifact therefore looks like Leadbay, not like a generic page.\n\nUse the tokens rather than hardcoded colours \u2014 the same rule the style package\nenforces. Re-theme by overriding them; don\'t fight specificity:\n\n```css\n:root { --lb-surface: var(--color-gray-2); --lb-radius: 0.5rem; }\n```\n\nDark mode works two ways: `data-theme="dark"` on `<html>` (the frontend\'s own\nhook) **and** `prefers-color-scheme`, because an artifact renders inside a host\nwhose theme attribute it cannot set. Never hardcode a light background over the\nskin.\n\nThe product face is `Nikkei Maru`; the stack names it first and falls back to\nthe system UI font. Do **not** add an `@font-face` \u2014 artifacts are inline-only\nand a remote font URL will silently fail.\n\n## What every lead card MUST carry\n\nA card is the artifact form of the `pull_leads` table, and it inherits that\ntable\'s rules. A card with a name and a button is not enough: the rep cannot\ntell *why* this lead is on screen. Four lines, in this order.\n\n```html\n<div class="lb-card">\n  <div class="lb-card-head">\n    <span class="lb-title"></span>              <!-- 1. company -->\n    <span class="lb-chips">                     <!-- 2. state -->\n      <span class="lb-chip" data-taste hidden></span>\n      <span class="lb-chip" data-status hidden></span>\n    </span>\n  </div>\n  <div class="lb-sub"></div>                    <!-- 3. firmographics -->\n  <div class="lb-sub" data-why></div>           <!-- 4. why it fits -->\n  <div class="lb-row"><!-- actions --></div>\n</div>\n```\n\n1. **Company** \u2014 `name`, linked to `website` (prefix `https://` on a bare host).\n   Never render the numeric `score`; use the `\u25B0\u2756\u25B1` bar if you want the signal.\n\n   Also give every card an **Open in Leadbay** link to the lead\'s panel in the\n   product. Put it at the **right-hand end of the card\'s last action row** \u2014\n   same row as the buttons, pushed right by an `lb-spacer`, not on a line of\n   its own. Style it `lb-link-out`: quiet text plus a plain arrow-up-right,\n   never a filled button. It is an escape hatch, not a call to action.\n\n   ```html\n   <div class="lb-row">\n     <button class="lb-btn">Like</button>\n     <button class="lb-btn">Set status</button>\n     <span class="lb-spacer"></span>          <!-- pushes the link right -->\n     <a class="lb-link-out" data-k="open" target="_blank" rel="noopener">\n       Open in Leadbay\n       <svg viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2"\n            stroke-linecap="round" stroke-linejoin="round" aria-hidden="true">\n         <line x1="7" y1="17" x2="17" y2="7"/><polyline points="7 7 17 7 17 17"/>\n       </svg>\n     </a>\n   </div>\n   ```\n\n   Keep the arrow a bare diagonal stroke \u2014 the text already says where the link\n   goes, so the glyph only has to mark "leaves this page". Mark the `<svg>`\n   `aria-hidden="true"`: it is decorative, and the link text is the accessible\n   name.\n   **Pick the view the lead actually lives in** \u2014 the URL is\n   `/app/<view>?lead=<uuid>`, and the three views are `discover`, `monitor`,\n   `campaign`. Landing a Monitor lead on Discover drops the rep into a list\n   that does not contain it:\n\n   ```js\n   function leadUrl(lead, campaignId) {\n     const id = encodeURIComponent(lead.id);\n     // A campaign card carries TWO params \u2014 the campaign selects the list, the\n     // lead opens the panel inside it. Campaign wins even when in_monitor is\n     // also true, because that is the list the rep is looking at.\n     if (campaignId) {\n       return `https://leadbay.app/app/campaign?campaign=${encodeURIComponent(campaignId)}&lead=${id}`;\n     }\n     const view = lead.in_monitor ? "monitor" : "discover";\n     return `https://leadbay.app/app/${view}?lead=${id}`;\n   }\n   openEl.href = leadUrl(lead, campaignId);\n   ```\n\n   `in_monitor` / `in_discover` are booleans on the `pull_followups` payload \u2014\n   every follow-up carries `in_monitor: true`, so a call sheet must link to\n   `monitor`. `pull_leads` omits both flags entirely; its leads are the Discover\n   batch by definition, so `discover` is the default. A campaign card\n   (`lb.callList({source:"campaign", campaignId})`) needs `?campaign=<id>&lead=<id>`\n   \u2014 the param names are `CAMPAIGN_QUERY_PARAM` and `LEAD_QUERY_PARAM`, and the\n   app\'s own `useLeadPanel` preserves whatever params are already set, so the\n   two coexist by design. Omitting `campaign=` opens an empty campaign view.\n\n   Inline the glyph as SVG rather than an emoji or `\u2197` \u2014 it inherits\n   `currentColor` and scales with the text, so it stays legible in both themes.\n   `?lead=<uuid>` is the real deep-link (`LEAD_QUERY_PARAM` in the web app, read\n   on load; the panel is an overlay, so the view choice only decides what sits\n   behind it). This is the ONE place a card may use `lead.id`: as a link target,\n   never as visible text.\n2. **State chips** \u2014 taste (`data-taste`) and CRM status (`data-status`) are\n   INDEPENDENT axes; render both, hide the empty one. Never collapse to one chip.\n3. **Firmographics** \u2014 sector of activity first, then city, then size, then the\n   contact. `sector_id` is a RAW ID (`"5136"`), not a label: resolve it via\n   `leadbay_list_sectors` (1346 rows \u2014 fetch once, cache, never inline the lot)\n   or omit it. Never print the raw id.\n\n   **Always show whether the lead is reachable \u2014 and never merge the person\n   with the company\'s switchboard.** These are two separate lines:\n\n   ```js\n   // WHO \u2014 recommended_contact. Name, and job_title ONLY when present; on list\n   // payloads it is usually null, and inventing one is worse than omitting it.\n   const rc = lead.recommended_contact;\n   const who = rc ? [rc.first_name, rc.last_name].filter(Boolean).join(" ") : null;\n   const whoLine = who ? who + (rc.job_title ? " \xB7 " + rc.job_title : "") : "No named contact";\n\n   // HOW \u2014 company-level channels. `phone_numbers` and `email` belong to the\n   // COMPANY, not to `recommended_contact`. Rendering "Jean \xB7 \u260E 0123\u2026" claims a\n   // direct line that does not exist; it is the switchboard.\n   const phone = (lead.phone_numbers || [])[0] || null;\n   // The API returns the STRING "null" for a missing email \u2014 guard for it or\n   // you will print the word "null" as an address.\n   const email = lead.email && lead.email !== "null" ? lead.email : null;\n   const howLine = [phone && "\u260E " + phone, email && "\u2709 " + email].filter(Boolean)\n     .join(" \xB7 ") || "No direct channel \u2014 enrich to reveal";\n   ```\n\n   ```html\n   <div class="lb-sub">Sector \xB7 City \xB7 Size</div>\n   <div class="lb-sub">\u{1F464} Jean-Fran\xE7ois Froemer \xB7 G\xE9rant</div>   <!-- WHO -->\n   <div class="lb-sub">\u{1F3E2} \u260E 01 23 45 67 89</div>                 <!-- HOW: company -->\n   ```\n\n   Label the channel line as the **company\'s**, so a rep reading fast cannot\n   mistake it for a direct line. A per-contact email or phone exists only after\n   enrichment \u2014 `research_lead_by_id` exposes it as `contacts.reachable[]`, and\n   `_meta.has_reachable_contact` is the authoritative flag. The list payloads\n   carry neither, so a card built from `pull_leads` / `pull_followups` can only\n   ever show company channels. Say "enrich to reveal" rather than implying the\n   contact is callable.\n\n   Two things that look like reachability and are not: a `linkedin_page` alone\n   (the rep cannot message a URL without leaving the artifact \u2014 same rule\n   `research_lead_by_id` applies), and `contacts_count > 0` (it counts known\n   people, not people you can contact; a lead can show 2518 contacts and zero\n   channels). `pull_followups` carries `has_phone` as a ready-made boolean;\n   `pull_leads` omits it, so derive from `phone_numbers` there.\n4. **Why it fits** \u2014 one sentence, \u226420 words. Walk this chain and stop at the\n   first hit:\n\n   1. `short_description`\n   2. `description` (longer; only on `research_lead_by_id` /\n      `research_lead_by_name_fuzzy` \u2014 the trim payloads omit it)\n   3. top 2 `tags[].display_name`\n   4. `qualification_summary.best_response_excerpt`, trimmed to one sentence\n   5. `keywords`, first 3, joined with ` \xB7 `\n   6. the resolved sector label \u2014 better than nothing, and if step 3 already\n      printed the sector on the firmographics line, skip to step 7\n   7. the literal *"No description yet \u2014 run qualification to generate one"*\n\n   Never leave this line blank: a silent gap reads as a rendering bug, whereas\n   the fallback tells the rep the data is missing and what fixes it.\n\n   **The two list payloads are complementary, so the chain must span both.**\n   `pull_leads` returns `short_description` on every lead but no `sector_id`;\n   `pull_followups` returns `sector_id` but no `short_description` at all. A\n   card fed by one will fall through to a different step than the same card fed\n   by the other \u2014 that is expected, not a bug. Never call\n   `research_lead_by_id` per row just to fill this line: it is one request per\n   lead. Fetch it lazily when the rep expands a card.\n\n**Never show** on a card: `id`, `sector_id`, `location.pos`, `location.country`\n(unless city and state are both missing), `is_hq`, `*_in_progress`,\n`highlighted_fields`, `custom_fields`, `stale_at`, `deal_insights`,\n`need_attention*`, any count that is 0, any value that is the string `"null"`.\n\n**Minimum actions.** A card that only displays is a table row that costs more \u2014\nif you are not wiring an action, render the markdown table instead. Wire at\nleast one write, and prefer the set the rep actually needs:\n\n| Card is for | Wire |\n|---|---|\n| triage a discovery batch | `lb.like` / `lb.dislike` + `lb.setStatus` |\n| working a call list | `lb.outreach` (gated on a note) + `lb.leadHistory` |\n| pipeline review | `lb.setStatus` + `lb.note` |\n\nAlways render the `.error` branch of every view-model \u2014 a control that cannot\nreach the host must say so, not sit silent.\n\n## Recipe: cold-call sheet (one row per lead)\n\n```js\nconst lb = window.LeadbayArtifacts; lb.configure();\nconst ASK = "<the user\'s request>";\n\nconst list = lb.callList({ source: "campaign", campaignId: CID, ask: ASK });\nlist.subscribe((l) => renderRows(l.items, l.loading));   // your render\n\n// per lead row (call when you build a row):\nfunction wireRow(lead, els) {\n  const status = lb.field({ value: "STILL_CHASING" });   // static-enum <select>\n  const note   = lb.field({ validate: (v) => (v && v.trim() ? null : "Add a note") });\n  lb.bindValue(els.status, status);\n  lb.bindValue(els.note, note);\n  lb.bindAction(els.log,  lb.outreach({ leadId: lead.id, ask: ASK, status, note }));\n  lb.bindAction(els.like, lb.like(lead.id));\n\n  const history = lb.leadHistory(lead.id, ASK);          // lazy\n  history.subscribe((h) => renderHistory(els.history, h));\n  els.expand.onclick = () => history.load();             // load on click\n}\n```\n\n## Recipe: lead-status dropdown (Wanted / Won / Lost)\n\nThe org-wide CRM status, as a `<select>` + Apply button. You write the markup;\n`lb.leadStatus` fills the options and holds the value, `lb.setStatus` does the write.\n\n```html\n<div class="lb-card">\n  <div class="lb-card-head">\n    <span class="lb-title">Acme Corp</span>\n    <span class="lb-chips">\n      <span id="taste" class="lb-chip" data-taste="liked">Liked</span>\n      <span id="crm"   class="lb-chip" data-status="WANTED">Wanted</span>\n    </span>\n  </div>\n  <div class="lb-row">\n    <select id="st" class="lb-select"></select>\n    <button id="go" class="lb-btn">Apply</button>\n    <span id="msg" class="lb-msg"></span>\n  </div>\n</div>\n```\n\n**Two badges, never one.** Taste (`liked`/`disliked`, from `lb.like`/`lb.dislike`)\nand CRM status (`WANTED`/`WON`/`LOST`/`UNWANTED`, from `lb.setStatus`) are\nindependent axes \u2014 a lead can be liked *and* lost. Collapsing them into a single\nchip destroys information: the rep can no longer see that a lead they liked went\nnowhere. Render `data-taste` and `data-status` as separate chips inside\n`lb-chips`, and hide the one that has no value rather than reusing it.\n\n```js\nlb.styles();                                          // once per artifact \u2014 see below\n\nconst status = lb.leadStatus(lead.org_lead_status);   // seed with the CURRENT value\nconst save   = lb.setStatus({ leadId: lead.id, status, ask: ASK });\n\nlb.bindSelect(document.getElementById("st"), status); // populates the 4 options\nlb.bindAction(document.getElementById("go"), save);   // click \u2192 write\n\nsave.subscribe((a) => {                               // render your own feedback\n  msg.textContent = a.loading ? "Saving\u2026"\n    : a.error ? a.error.message                       // includes partial failures\n    : a.lastResult ? `Set to ${a.lastResult.status}` : "";\n  msg.dataset.tone = a.error ? "error" : a.lastResult ? "ok" : "";\n});\n```\n\nLoading / success / error styling comes free: `bindAction` and `bindSelect` set\n`data-lb-state` (`ready|loading|error|success|unavailable`) and the skin already\ntargets those attributes. No extra wiring.\n\nSave-on-change instead of an Apply button \u2014 drop `bindAction` and run it yourself:\n\n```js\ndocument.getElementById("st").addEventListener("change", () => save.run());\n```\n\n**Bulk apply** across checked rows \u2014 pass `leadIds` and a `confirm`, since one\nclick rewrites a field every rep in the org sees:\n\n```js\nconst bulk = lb.setStatus({\n  leadIds: () => checkedIds,        // \u2190 read at run() time, not at build time\n  status, ask: ASK,\n  confirm: "Set this status on every selected lead?",\n});\n```\n\n`leadIds` is read when the action runs, so a live selection works \u2014 but pass the\narray itself if your selection is fixed. A partial write (some leads rejected)\nsurfaces as `.error`, never as a green button: `setStatus` checks the `failed[]`\nthe tool returns.\n\nThe backend stamps the status date as "now" on every write, which is what a rep\nclicking a dropdown means. Don\'t add a date picker unless the user asks to\nbackdate \u2014 then pass an optional `date` field holding `YYYY-MM-DD`:\n`lb.setStatus({ leadId, status, date, ask })`.\n\n## Recipe: manager dashboard\n\n```js\nconst team = lb.teamActivity({ weeks: 4, ask: ASK });\nteam.subscribe((t) => {\n  if (t.loading) showSpinner();\n  if (t.data) {\n    renderLeaderboard(t.data.reps);   // sorted by total_activities; cols: name, notes, meetings_or_interest, lost\u2026\n    renderTrendChart(t.data.trend);   // [{date,count}] \u2192 Chart.js (allowed from CDN)\n  }\n});\nrefreshBtn.onclick = () => team.refresh();\n```\n\n## Recipe: live enrichment\n\n```js\nconst job = lb.enrichment({ leadIds: [LEAD], titles: ["CEO", "VP Sales"], ask: ASK });\njob.subscribe((j) => {\n  const p = j.data && j.data.overall_progress;            // {done,total,done_ratio}\n  renderBar(p);\n  if (j.done) renderContacts(j.data.leads);               // enriched contacts\n});\nrefreshBtn.onclick = () => job.refresh();\n```\n\n## Write-call rules\n\nThe domain factories handle these for you. If you hand-roll an action:\n`leadbay_report_outreach` args MUST include `verification:{source:"user_confirmed", ref}`\nAND `_triggered_by`; `leadbay_add_leads_to_campaign` needs `_triggered_by`;\n`add_note`/`like_lead`/`dislike_lead` take only their own args. `epilogue_status` is\none of `lb.EPILOGUE_STATUSES`. Snoozing (pushback) is advanced-gated \u2014 not\ncallable from a default artifact. Org lead status IS on the default surface:\nuse `lb.setStatus`, which owns the arg shape AND the partial-write check \u2014\n`leadbay_set_lead_status` writes each lead separately, so it can resolve 200\nwith a non-empty `failed[]`. Hand-rolling that action will report a green\nbutton over a write that never landed.\n\n## Degradation + live updates\n\nIf the host bridge is absent, a view-model\'s `.error` is set with `.error.unavailable\n=== true` (bind helpers set `data-lb-state="unavailable"`) \u2014 nothing throws. Every\ncall also has a **30s timeout** (configurable via `lb.configure({ timeoutMs })`): a\nhost call that never settles becomes `.error` with `code:"timeout"`, so a control is\nnever stuck loading forever \u2014 always render the `.error` branch so the user can retry.\nAuto-poll (`pollEvery`) depends on the cowork host serving FRESH reads; `.refresh()`\nis the guaranteed manual path \u2014 always wire a Refresh control for polling resources.';
+var ARTIFACT_RUNTIME = '"use strict";(()=>{var E=Object.defineProperty;var A=(e,t,r)=>t in e?E(e,t,{enumerable:!0,configurable:!0,writable:!0,value:r}):e[t]=r;var a=(e,t,r)=>A(e,typeof t!="symbol"?t+"":t,r);var S=`\n:root{\n--color-black:#191919;--color-white:#fff;\n--color-gray-1:#f9f9f9;--color-gray-2:#f0f0f0;--color-gray-3:#e0e0e0;--color-gray-4:#cecece;\n--color-gray-5:#c4c4c4;--color-gray-6:#8d8d8d;--color-gray-7:#787878;--color-gray-8:#646464;\n--color-gray-9:#202020;\n--color-linkedin:#0a66c2;\n--color-blue-background:oklch(0.947 0.029 251);--color-blue-foreground:oklch(0.564 0.181 251);\n--color-green-background:oklch(0.947 0.029 141);--color-green-foreground:oklch(0.564 0.181 141);\n--color-red-background:oklch(0.947 0.029 26);--color-red-foreground:oklch(0.564 0.191 26);\n--color-gold-background:oklch(0.972 0.049 91);--color-gold-foreground:oklch(0.667 0.177 91);\n--color-cherry-background:oklch(0.947 0.029 15);--color-cherry-foreground:oklch(0.44 0.146 15);\n--color-red-like:var(--color-cherry-foreground);\n--lb-font:"Nikkei Maru",system-ui,-apple-system,"Segoe UI",Roboto,sans-serif;\n/* Concentric: outer = inner + padding (0.625 + 0.875 = 1.5rem). At 1rem the\n   inner control\'s arc outran the card\'s in the corner. */\n--lb-radius:1.5rem;--lb-radius-sm:0.625rem;--lb-gap:0.75rem;\n--lb-surface:var(--color-gray-1);--lb-border:var(--color-gray-3);\n--lb-fg:var(--color-black);--lb-muted:var(--color-gray-8);--lb-field:var(--color-white);\n/* Chips and shadows need their own aliases: a chip painted with a raw ramp\n   value cannot theme, and a shadow tinted with --color-gray-9 is invisible in\n   dark because gray-9 IS the dark surface. */\n--lb-chip-bg:var(--color-gray-2);\n--lb-control-border:var(--color-gray-6);\n--lb-shadow:0 0.5rem 1.5rem oklch(0 0 0/.06),0 0.125rem 0.5rem oklch(0 0 0/.04);\ncolor-scheme:light;\n}\n:root[data-theme=dark],:root[data-lb-theme=dark]{\n--lb-surface:var(--color-gray-9);--lb-border:var(--color-gray-8);\n--lb-fg:var(--color-white);--lb-muted:var(--color-gray-3);\n/* --lb-field must NOT equal --lb-surface: a control needs its own ground or it\n   reads as text on the card (was 1.00:1). */\n--lb-field:#2c2c2c;\n--lb-chip-bg:#3b3b3b;\n--lb-control-border:var(--color-gray-6);\n--lb-shadow:0 0.5rem 1.5rem oklch(0 0 0/.5),0 0.125rem 0.5rem oklch(0 0 0/.35);\ncolor-scheme:dark;\n--color-blue-background:oklch(0.30 0.055 251);--color-blue-foreground:oklch(0.80 0.11 251);\n--color-green-background:oklch(0.30 0.055 141);--color-green-foreground:oklch(0.80 0.13 141);\n--color-red-background:oklch(0.30 0.060 26);--color-red-foreground:oklch(0.80 0.11 26);\n--color-gold-background:oklch(0.30 0.055 91);--color-gold-foreground:oklch(0.82 0.13 91);\n--color-cherry-background:oklch(0.30 0.055 15);--color-cherry-foreground:oklch(0.82 0.10 15);\n}\n@media(prefers-color-scheme:dark){:root:not([data-theme=light]):not([data-lb-theme=light]){\n--lb-surface:var(--color-gray-9);--lb-border:var(--color-gray-8);\n--lb-fg:var(--color-white);--lb-muted:var(--color-gray-3);\n/* --lb-field must NOT equal --lb-surface: a control needs its own ground or it\n   reads as text on the card (was 1.00:1). */\n--lb-field:#2c2c2c;\n--lb-chip-bg:#3b3b3b;\n--lb-control-border:var(--color-gray-6);\n--lb-shadow:0 0.5rem 1.5rem oklch(0 0 0/.5),0 0.125rem 0.5rem oklch(0 0 0/.35);\ncolor-scheme:dark;\n--color-blue-background:oklch(0.30 0.055 251);--color-blue-foreground:oklch(0.80 0.11 251);\n--color-green-background:oklch(0.30 0.055 141);--color-green-foreground:oklch(0.80 0.13 141);\n--color-red-background:oklch(0.30 0.060 26);--color-red-foreground:oklch(0.80 0.11 26);\n--color-gold-background:oklch(0.30 0.055 91);--color-gold-foreground:oklch(0.82 0.13 91);\n--color-cherry-background:oklch(0.30 0.055 15);--color-cherry-foreground:oklch(0.82 0.10 15);\n}}\n.lb-card{display:grid;gap:var(--lb-gap);padding:0.875rem;\nbackground-color:var(--lb-surface);border:1px solid var(--lb-border);\nborder-radius:var(--lb-radius);corner-shape:squircle;color:var(--lb-fg);\nfont-family:var(--lb-font);\nbox-shadow:var(--lb-shadow)}\n/* The head must be able to wrap. Both children default to flex:0 1 auto, so\n   without this they shrink together and a long company name is crushed to three\n   lines while the chips keep full width. flex-basis:11rem is a content breakpoint:\n   below ~176px for the title, the chips drop to their own line instead. */\n.lb-card-head{display:flex;flex-wrap:wrap;justify-content:space-between;\nalign-items:baseline;gap:var(--lb-gap)}\n/* An email or URL has no break opportunity \\u2014 browsers do not break at "@" or\n   ".", so without this a company address escapes the card and the chat host\n   clips it with no way to recover the tail. */\n.lb-title{font-size:0.875rem;font-weight:600;line-height:1.3;color:var(--lb-fg);\noverflow-wrap:break-word;min-width:0;flex:1 1 11rem}\n.lb-sub{font-size:0.8125rem;line-height:1.45;color:var(--lb-muted);\noverflow-wrap:break-word;max-width:68ch}\n/* An empty fact line is a 30px void that reads as a rendering bug. */\n.lb-sub:empty{display:none}\n/* Every card child shared one 12px gap, so nothing grouped: the switchboard\n   number, the contact and the fit judgement read as one paragraph. 4px inside\n   the facts vs 12px to the why line is a 3x ratio \\u2014 grouping by space, with no\n   rule or border. */\n.lb-facts{display:grid;gap:0.25rem}\n/* The contract\'s canonical skeleton already emits data-why; the sheet just did\n   not honour it, so consumers hand-patched an inline font-style. */\n.lb-sub[data-why]{font-style:italic;color:var(--lb-fg)}\n/* Icon column: the glyph gets its own cell so a wrapping line hangs on one\n   shared text edge instead of running back under the emoji. */\n.lb-fact{display:grid;grid-template-columns:1.15rem 1fr;gap:0.35rem;align-items:baseline}\n.lb-row{display:flex;align-items:center;gap:var(--lb-gap);flex-wrap:wrap}\n/* row-gap matters once the row wraps: without it the wrapped line sits flush\n   against the one above. */\n.lb-row{row-gap:0.5rem}\n.lb-stack{display:grid;gap:var(--lb-gap)}\n.lb-select,.lb-input{font:inherit;font-family:var(--lb-font);font-size:0.8125rem;color:var(--lb-fg);\nbackground-color:var(--lb-field);border:1px solid var(--lb-control-border);\nborder-radius:var(--lb-radius-sm);corner-shape:squircle;padding:0.4rem 0.55rem;min-height:2.125rem}\n.lb-btn{font:inherit;font-family:var(--lb-font);font-size:0.8125rem;font-weight:600;\ncolor:var(--lb-fg);background-color:var(--lb-field);border:1px solid var(--lb-control-border);\nborder-radius:var(--lb-radius-sm);corner-shape:squircle;padding:0.4rem 0.85rem;min-height:2.125rem;\ncursor:pointer;transition:background-color .15s,border-color .15s,color .15s}\n.lb-btn:hover:not([disabled]){border-color:var(--color-gray-7)}\n/* A SUBMIT is not a toggle. Like/Dislike are independent switches; "Set status"\n   commits the value sitting in the select beside it. Rendered identically, the\n   row reads as five peers and the rep cannot see which control ends the task.\n   The accent border + tinted ground marks it as the commit without making it a\n   loud filled button \\u2014 the state styles above still win when they apply. */\n/* INVERTED, not recoloured. A normal button is --lb-fg on --lb-field; the\n   submit swaps them. That introduces no new hue, and it themes for free: light\n   gives near-black on white, dark gives white on near-black \\u2014 the polarity flip\n   reads as "this one commits" in both, without competing with the semantic\n   state colours (success green / error red) that still override it. */\n.lb-btn-submit{background-color:var(--lb-fg);color:var(--lb-field);\nborder-color:var(--lb-fg)}\n.lb-btn-submit:hover:not([disabled]){\nbackground-color:color-mix(in srgb,var(--lb-fg) 85%,var(--lb-field));\nborder-color:var(--lb-fg)}\n/* Toggles that act on the same axis belong in one visual unit, so the row reads\n   as "taste | status | inspect" rather than five equal buttons. The gap inside\n   a group is half the gap between groups \\u2014 the 2x rule, applied horizontally. */\n.lb-group{display:inline-flex;align-items:center;gap:0.375rem;flex-wrap:wrap}\n/* Square icon button. Only for actions whose glyph is unambiguous on its own \\u2014\n   thumbs up/down qualify, "Set status" does not. It MUST carry an aria-label\n   and a title: with no text the icon is the entire affordance, so an unlabelled\n   one is unreadable to a screen reader and to anyone who does not know the\n   glyph. Same 2.125rem box as a text button, so it keeps the row\'s baseline. */\n.lb-btn-icon{padding:0;width:2.125rem;justify-content:center;display:inline-flex;\nalign-items:center}\n.lb-btn-icon svg{width:1.05rem;height:1.05rem;flex-shrink:0}\n/* Pressed state. The taste chip carries the durable fact; this is the control\n   reflecting it, and it must not rely on colour alone \\u2014 aria-pressed is the\n   accessible half and the fill is the visible half. */\n/* A heart that stays hollow does not read as "liked" \\u2014 fill it from the same\n   currentColor the stroke uses, so one rule covers both themes. */\n.lb-btn-icon[aria-pressed=true] svg{fill:currentColor}\n.lb-btn-icon[aria-pressed=true]{border-color:var(--color-red-like);\ncolor:var(--color-red-like);\nbackground-color:color-mix(in srgb,var(--color-red-like) 10%,var(--lb-field))}\n.lb-btn-icon[data-taste=disliked][aria-pressed=true]{border-color:var(--lb-muted);\ncolor:var(--lb-muted);\nbackground-color:color-mix(in srgb,var(--lb-muted) 10%,var(--lb-field))}\n/* A glyph-only control has no text to widen its hit area, so extend the target\n   past the 34px box without inflating the row. */\n.lb-btn-icon{position:relative}\n.lb-btn-icon::before{content:"";position:absolute;inset:-5px}\n.lb-btn:focus-visible,.lb-select:focus-visible,.lb-input:focus-visible,\n.lb-link:focus-visible{\noutline:2px solid var(--color-blue-foreground);outline-offset:1px;\nborder-radius:var(--lb-radius-sm)}\n/* Forced colours replaces every background and custom outline with a system\n   colour, collapsing the state styling above into one flat control. */\n@media(forced-colors:active){\n.lb-btn,.lb-select,.lb-input,.lb-chip{border:1px solid CanvasText}\n.lb-btn:focus-visible,.lb-select:focus-visible,.lb-input:focus-visible,\n.lb-link:focus-visible,.lb-link-out:focus-visible{outline:2px solid Highlight}\n.lb-link-out{opacity:1}}\n/* Each state carries a glyph as well as a hue: success and error were two\n   pills with identical labels, indistinguishable to a red-green colour-blind\n   reader and identical under forced colours. */\n.lb-btn[data-lb-state=loading]{opacity:.55;cursor:progress}\n.lb-btn[data-lb-state=loading]::after{content:"\\u2026";margin-inline-start:.35rem}\n/* An icon button has no room for a trailing glyph \\u2014 it would push the icon\n   off-centre. Its states read through border and fill, which the forced-colors\n   block below keeps distinguishable. */\n.lb-btn-icon::after{content:none!important}\n.lb-btn[data-lb-state=success]::after{content:"\\u2713";margin-inline-start:.35rem}\n.lb-btn[data-lb-state=error]::after{content:"\\u2715";margin-inline-start:.35rem}\n.lb-btn[data-lb-state=unavailable]::after{content:"\\u26A0";margin-inline-start:.35rem}\n/* With save-on-change there is no button to reflect the write, so the select\n   itself must. Only error was styled before; loading and success were invisible\n   on a select, leaving the rep unsure whether the pick had landed. */\n.lb-select[data-lb-state=loading]{opacity:.55;cursor:progress}\n.lb-select[data-lb-state=success]{border-color:var(--color-green-foreground)}\n.lb-msg[data-tone=error]::before{content:"\\u2715 "}\n.lb-msg[data-tone=ok]::before{content:"\\u2713 "}\n.lb-btn[data-lb-state=success]{background-color:var(--color-green-background);\nborder-color:var(--color-green-foreground);color:var(--color-green-foreground)}\n.lb-btn[data-lb-state=error],.lb-select[data-lb-state=error]{\nbackground-color:var(--color-red-background);border-color:var(--color-red-foreground);\ncolor:var(--color-red-foreground)}\n.lb-btn[data-lb-state=unavailable],.lb-btn[disabled]{opacity:.5;cursor:not-allowed}\n.lb-msg{font-size:0.8125rem;line-height:1.45;color:var(--lb-muted)}\n/* Screen-reader-only text: control labels, unit suffixes, "opens in a new tab". */\n.lb-vh{position:absolute;width:1px;height:1px;margin:-1px;padding:0;\noverflow:hidden;clip-path:inset(50%);white-space:nowrap;border:0}\n.lb-msg[data-tone=error]{color:var(--color-red-foreground)}\n.lb-msg[data-tone=ok]{color:var(--color-green-foreground)}\n.lb-chip{display:inline-flex;align-items:center;gap:.25rem;white-space:nowrap;\nfont-size:0.75rem;font-weight:600;line-height:1rem;padding:0.125rem 0.5rem;\nborder-radius:var(--lb-radius-sm);corner-shape:squircle;\nbackground-color:var(--lb-chip-bg);color:var(--lb-muted)}\n.lb-chip[data-status=WANTED]{background-color:var(--color-blue-background);color:var(--color-blue-foreground)}\n.lb-chip[data-status=WON]{background-color:var(--color-green-background);color:var(--color-green-foreground)}\n.lb-chip[data-status=LOST]{background-color:var(--color-red-background);color:var(--color-red-foreground)}\n.lb-chip[data-status=UNWANTED]{background-color:var(--lb-chip-bg);color:var(--lb-muted)}\n.lb-chip[data-taste=liked]{background-color:var(--color-cherry-background);color:var(--color-red-like)}\n.lb-chip[data-taste=disliked]{background-color:var(--lb-chip-bg);color:var(--lb-muted)}\n.lb-chips{display:flex;align-items:center;gap:.35rem;flex-wrap:wrap}\n.lb-chip[hidden]{display:none}\n.lb-table{width:100%;border-collapse:collapse;font-family:var(--lb-font);color:var(--lb-fg)}\n.lb-table th,.lb-table td{text-align:start;padding:0.5rem 0.4rem;\nborder-bottom:1px solid var(--lb-border);vertical-align:middle;font-size:0.8125rem}\n.lb-table th{font-size:0.75rem;font-weight:600;color:var(--lb-muted);\ntext-transform:uppercase;letter-spacing:.04em}\n.lb-link{color:var(--color-blue-foreground);text-decoration:none}\n.lb-link:hover{text-decoration:underline}\n/* Quiet text link out of the artifact. Button-height so it shares the row\'s\n   baseline; understated so it never competes with the actions beside it. The\n   arrow is a bare diagonal stroke \\u2014 an escape-hatch marker, not an icon that\n   asks to be read. */\n.lb-link-out{display:inline-flex;align-items:center;gap:.3rem;\nfont-size:0.75rem;line-height:1rem;min-height:2.125rem;\ncolor:var(--lb-fg);text-decoration:none;opacity:.65;transition:opacity .15s}\n.lb-link-out:hover{opacity:1;text-decoration:underline}\n.lb-link-out:focus-visible{outline:2px solid var(--color-blue-foreground);outline-offset:1px;\nborder-radius:var(--lb-radius-sm)}\n.lb-link-out svg{width:.85em;height:.85em;flex-shrink:0}\n/* Pushes whatever follows it to the right edge of an .lb-row, so a trailing\n   link sits on the SAME baseline as the row\'s buttons instead of on its own\n   line below them. */\n/* A spacer only aligns items on its OWN flex line. .lb-row wraps, so once the\n   link falls to a second line the spacer stays behind and the link renders\n   left-aligned \\u2014 the opposite of what the contract promises. The auto margin\n   re-anchors it to the trailing edge of whichever line it lands on, and is a\n   no-op while the row is unwrapped. */\n.lb-spacer{flex:1 1 auto;min-width:0}\n.lb-row>.lb-link-out{margin-inline-start:auto}\n.lb-spinner{display:inline-block;width:.7em;height:.7em;border:2px solid var(--lb-border);\nborder-top-color:var(--color-blue-foreground);border-radius:50%;animation:lb-spin .8s linear infinite}\n@keyframes lb-spin{to{transform:rotate(1turn)}}\n/* animation:none left a static ring \\u2014 under reduced motion "sending" and\n   "cannot click" looked identical. An opacity pulse is the sanctioned\n   substitute. The 150ms colour fade is fine and stays. */\n@media(prefers-reduced-motion:reduce){.lb-spinner{animation:lb-pulse 1.4s ease-in-out infinite}}\n@keyframes lb-pulse{50%{opacity:.35}}\n`,v="lb-styles";var L="0.5.0",u=class extends Error{constructor(r,o={}){super(r);a(this,"code");a(this,"raw");this.name="LbError",this.code=o.code,this.raw=o.raw}},k=null,h=3e4;function C(){let e=globalThis.cowork;return e&&typeof e.callMcpTool=="function"?(t,r)=>e.callMcpTool(t,r):null}function T(e){if(e&&typeof e=="object"&&"content"in e){let t=e.content;if(Array.isArray(t)&&t[0]&&typeof t[0].text=="string")return t[0].text}return null}function _(e){if(!e||typeof e!="object")return e;let t=e;if(t.isError)throw new u(T(e)??"tool call failed",{raw:e});if("structuredContent"in t&&t.structuredContent!=null)return t.structuredContent;let r=T(e);if(r!=null)try{return JSON.parse(r)}catch{return r}return e}function I(e){return e instanceof Error?e.message:String(e)}function p(e){let t=e instanceof u?e.code:void 0;return{message:I(e),unavailable:t==="unavailable",code:t}}function O(e={}){k=e.call??null,h=e.timeoutMs??3e4}function R(){if(typeof document>"u"||!document.head)return null;let e=document.getElementById(v);if(e)return e;let t=document.createElement("style");return t.id=v,t.textContent=S,document.head.appendChild(t),t}async function x(e,t){if(!h||h<=0)return e;let r,o=new Promise((n,i)=>{r=setTimeout(()=>i(new u(`"${t}" timed out after ${h}ms`,{code:"timeout"})),h)});try{return await Promise.race([e,o])}finally{r&&clearTimeout(r)}}async function l(e,t={}){if(k)return _(await x(Promise.resolve(k(e,t)),e));let r=C();if(!r)throw new u("Leadbay bridge unavailable (window.cowork absent)",{code:"unavailable"});return _(await x(Promise.resolve(r(e,t)),e))}var b=class{constructor(){a(this,"subs",new Set)}subscribe(t){return this.subs.add(t),t(this),()=>this.subs.delete(t)}emit(){for(let t of this.subs)t(this)}};function N(e){return Array.isArray(e)?e.map(t=>t&&typeof t=="object"?t:{value:t,label:String(t)}):[]}var g=class extends b{constructor(r={}){super();a(this,"kind");a(this,"value");a(this,"options",[]);a(this,"loading",!1);a(this,"error",null);a(this,"ready",!1);a(this,"cfg");a(this,"depUnsubs",[]);a(this,"seq",0);this.cfg=r,this.kind=r.kind,this.value=r.value??"";for(let o of r.dependsOn??[]){let n=o.value;this.depUnsubs.push(o.subscribe(()=>{o.value!==n&&(n=o.value,this.cfg.load&&this.load())}))}r.load&&(r.autoLoad??!0)&&this.load()}async load(){if(!this.cfg.load)return;let r=++this.seq;this.loading=!0,this.error=null,this.emit();try{let o=await this.cfg.load();if(r!==this.seq)return;this.options=this.cfg.options?this.cfg.options(o):N(o),this.ready=!0;let n=this.value==null?"":String(this.value);this.options.length&&(n===""||!this.options.some(i=>String(i.value)===n))&&(this.value=this.options[0].value)}catch(o){if(r!==this.seq)return;this.options=[],this.error=p(o)}finally{r===this.seq&&(this.loading=!1,this.emit())}}setValue(r){this.value=r;let o=this.validate();this.error=o?{message:o,unavailable:!1}:null,this.emit()}validate(){return this.cfg.validate?this.cfg.validate(this.value):null}get valid(){return this.validate()==null}reset(){this.value=this.cfg.value??"",this.error=null,this.emit()}dispose(){for(let r of this.depUnsubs)r();this.depUnsubs=[]}};function P(e){if(!e||typeof e!="object")return null;let t=e;if(t.error!==!0)return null;let r=typeof t.message=="string"&&t.message?t.message:"tool call failed",o=typeof t.hint=="string"&&t.hint?` \\u2014 ${t.hint}`:"";return`${r}${o}`}var s=class extends b{constructor(r){super();a(this,"loading",!1);a(this,"error",null);a(this,"lastResult",null);a(this,"cfg");this.cfg=r}async run(){if(this.loading)return;for(let n of this.cfg.fields??[]){let i=n.validate();if(i!=null){this.error={message:i,unavailable:!1},this.emit();return}}if(this.cfg.confirm&&typeof globalThis.confirm=="function"&&!globalThis.confirm(this.cfg.confirm))return;this.loading=!0,this.error=null,this.emit();let r;try{let n=typeof this.cfg.args=="function"?this.cfg.args():this.cfg.args??{};r=await l(this.cfg.tool,n)}catch(n){this.error=p(n),this.loading=!1,this.emit(),this.cfg.onError?.(this.error);return}let o=P(r)??this.cfg.checkResult?.(r)??null;if(o!=null){this.error={message:o,unavailable:!1},this.loading=!1,this.emit(),this.cfg.onError?.(this.error);return}return this.lastResult=r,this.loading=!1,this.emit(),this.cfg.onSuccess?.(r),r}reset(){this.error=null,this.lastResult=null,this.emit()}},d=class extends b{constructor(r){super();a(this,"data",null);a(this,"loading",!1);a(this,"refreshing",!1);a(this,"error",null);a(this,"done",!1);a(this,"cfg");a(this,"timer",null);a(this,"seq",0);this.cfg=r,(r.autoLoad??!0)&&this.load()}async load(){this.clearTimer();let r=++this.seq;this.data==null?this.loading=!0:this.refreshing=!0,this.error=null,this.emit();try{let n=await this.cfg.load();if(r!==this.seq)return;this.data=n,this.done=this.cfg.until?this.cfg.until(n):!0,this.cfg.pollEvery&&!this.done&&(this.timer=setTimeout(()=>void this.load(),this.cfg.pollEvery))}catch(n){if(r!==this.seq)return;this.error=p(n)}finally{r===this.seq&&(this.loading=!1,this.refreshing=!1,this.emit())}}refresh(){return this.load()}stop(){this.clearTimer()}clearTimer(){this.timer&&(clearTimeout(this.timer),this.timer=null)}},f=class extends b{constructor(r){super();a(this,"items",[]);a(this,"page",0);a(this,"pageSize");a(this,"total",0);a(this,"loading",!1);a(this,"error",null);a(this,"cfg");a(this,"seq",0);this.cfg=r,this.pageSize=r.pageSize??20,(r.autoLoad??!0)&&this.loadPage(0)}async loadPage(r){let o=++this.seq;this.loading=!0,this.error=null,this.emit();try{let n=await this.cfg.load({page:r,pageSize:this.pageSize});if(o!==this.seq)return;this.items=n.items??[],this.total=n.total??this.items.length,this.page=r}catch(n){if(o!==this.seq)return;this.error=p(n)}finally{o===this.seq&&(this.loading=!1,this.emit())}}next(){return this.loadPage(this.page+1)}prev(){return this.loadPage(Math.max(0,this.page-1))}get hasMore(){return(this.page+1)*this.pageSize<this.total}};function M(e,t){let r=t.error?.unavailable?"unavailable":t.loading?"loading":t.error?"error":"ready";e.setAttribute("data-lb-state",r),t.error?e.setAttribute("data-lb-error",t.error.message):e.removeAttribute("data-lb-error")}function D(e,t){let r=()=>t.setValue(e.value);e.addEventListener("change",r);let o=t.subscribe(()=>{M(e,t),e.disabled=t.loading,e.innerHTML="";for(let n of t.options){let i=document.createElement("option");i.value=String(n.value),i.textContent=n.label,e.appendChild(i)}e.value=t.value==null?"":String(t.value)});return()=>{e.removeEventListener("change",r),o()}}function z(e,t){let r=e.type==="checkbox",o=e.tagName==="SELECT"?"change":"input",n=()=>t.setValue(r?e.checked:e.value);e.addEventListener(o,n);let i=t.subscribe(()=>{if(r)e.checked=!!t.value;else{let c=t.value==null?"":String(t.value);e.value!==c&&(e.value=c)}e.setAttribute("data-lb-state",t.error?"error":"ready"),t.error?e.setAttribute("data-lb-error",t.error.message):e.removeAttribute("data-lb-error")});return()=>{e.removeEventListener(o,n),i()}}function q(e,t){let r=i=>{i.preventDefault(),t.run()};e.addEventListener("click",r);let o,n=t.subscribe(()=>{let i=t.error?.unavailable?"unavailable":t.loading?"loading":t.error?"error":t.lastResult!=null?"success":"ready";e.setAttribute("data-lb-state",i),"disabled"in e&&(e.disabled=t.loading),e.setAttribute("aria-busy",String(t.loading)),t.error?e.setAttribute("data-lb-error",t.error.message):e.removeAttribute("data-lb-error"),o&&clearTimeout(o),i==="success"&&(o=setTimeout(()=>{!t.loading&&!t.error&&e.setAttribute("data-lb-state","ready")},1600))});return()=>{e.removeEventListener("click",r),n(),o&&clearTimeout(o)}}var F=["STILL_CHASING","COULD_NOT_REACH_STILL_TRYING","INTEREST_VALIDATED_OR_MEETING_PLANED","NOT_INTERESTED_LOST"],y=[{value:"",label:"Default ranking"},{value:"SCORE:DESC",label:"Score \\u2193"},{value:"SCORE:ASC",label:"Score \\u2191"},{value:"NAME:ASC",label:"Name A\\u2192Z"},{value:"NAME:DESC",label:"Name Z\\u2192A"},{value:"SIZE:DESC",label:"Size \\u2193"},{value:"SIZE:ASC",label:"Size \\u2191"},{value:"SECTOR:ASC",label:"Sector A\\u2192Z"},{value:"STATUS:ASC",label:"Status A\\u2192Z"},{value:"CONTACT_COUNT:DESC",label:"Contacts \\u2193"},{value:"LAST_PROSPECTING_ACTION_AT:DESC",label:"Last action \\u2193"},{value:"LAST_PROSPECTING_ACTION_AT:ASC",label:"Last action \\u2191"},{value:"EPILOGUE_STATUS_SET_AT:DESC",label:"Outcome set \\u2193"},{value:"LIKED:DESC",label:"Liked first"},{value:"DISLIKED:DESC",label:"Disliked first"}];function U(e){let t=String(e??"").trim().toUpperCase(),r=y.some(o=>o.value===t);return new g({kind:"select",value:r?t:"",load:async()=>y.slice()})}var m=[{value:"WANTED",label:"Wanted"},{value:"WON",label:"Won"},{value:"LOST",label:"Lost"},{value:"UNWANTED",label:"Unwanted"}],H={value:"",label:"Pick a status"};function j(e){let t=String(e??"").trim().toUpperCase(),r=m.some(o=>o.value===t);return new g({kind:"select",value:r?t:"",validate:o=>String(o??"")===""?"Pick a status":null,load:async()=>r?m.slice():[H,...m]})}function W(e){let t=()=>{let r=typeof e.leadIds=="function"?e.leadIds():e.leadIds;return Array.isArray(r)?r:e.leadId?[e.leadId]:[]};return new s({tool:"leadbay_set_lead_status",fields:e.date?[e.status,e.date]:[e.status],confirm:e.confirm,args:()=>({lead_ids:t(),status:e.status.value,...e.date&&e.date.value?{status_date:e.date.value}:{},...e.ask?{_triggered_by:e.ask}:{}}),checkResult:r=>{let o=r?.failed;if(!Array.isArray(o)||o.length===0)return null;let n=t().length,i=o[0]?.message??"write rejected";return o.length===n?`Status not applied: ${i}`:`${o.length} of ${n} leads failed: ${i}`}})}function V(e){return new g({kind:"select",load:()=>l("leadbay_list_campaigns",{_triggered_by:e}),options:t=>(t?.campaigns??[]).map(o=>{let n=o?.campaign??o;return n?.id?{value:n.id,label:n.name??n.ai_generated_name??String(n.id)}:null}).filter(o=>o!=null)})}function $(e){return new s({tool:"leadbay_report_outreach",fields:e.note?[e.note]:[],args:()=>({lead_id:e.leadId,...e.status?{epilogue_status:e.status.value}:{},note:e.note?e.note.value:"",verification:{source:"user_confirmed",ref:e.ref??"logged from artifact"},_triggered_by:e.ask})})}function G(e){return new s({tool:"leadbay_add_note",fields:[e.note],args:()=>({leadId:e.leadId,note:e.note.value})})}function B(e){return new s({tool:"leadbay_like_lead",args:{lead_id:e}})}function Z(e){return new s({tool:"leadbay_dislike_lead",args:{lead_id:e}})}function Y(e,t){return new d({autoLoad:!1,load:()=>l("leadbay_account_history",{leadId:e,_triggered_by:t})})}function K(e,t){return new d({autoLoad:!1,load:()=>l("leadbay_research_lead_by_id",{leadId:e,_triggered_by:t})})}function J(e){let t=null;return new d({...e.autoLoad!==void 0?{autoLoad:e.autoLoad}:{},pollEvery:e.pollEvery??4e3,until:r=>!!r?.all_done,load:async()=>{if(!t){let r=await l("leadbay_enrich_titles",{...e.leadIds?{leadIds:e.leadIds}:{},titles:e.titles,...e.email!==void 0?{email:e.email}:{},...e.phone!==void 0?{phone:e.phone}:{},...e.confirm!==void 0?{confirm:e.confirm}:{},_triggered_by:e.ask}),o=Array.isArray(r?.lead_ids)?r.lead_ids:[],n=r?.notification_id??null;if(t=n||o.length>0?{notification_id:n,lead_ids:o}:null,!t)return{...r,all_done:!0,no_job:!0}}return l("leadbay_bulk_enrich_status",{...t.notification_id?{notification_id:t.notification_id}:{},...t.lead_ids.length>0?{lead_ids:t.lead_ids}:{},...e.titles?{titles:e.titles}:{},...e.email!==void 0?{email:e.email}:{},...e.phone!==void 0?{phone:e.phone}:{},_triggered_by:e.ask})}})}function Q(e){let t=()=>typeof e.order=="string"?e.order:String(e.order?.value??"");return new f({pageSize:e.pageSize??20,load:async({page:r,pageSize:o})=>{let n=await l("leadbay_pull_leads",{page:r,count:o,...e.lensId?{lensId:e.lensId}:{},...t()?{order:t()}:{},_triggered_by:e.ask}),i=n.leads??[];return{items:i,total:n.pagination?.total??i.length}}})}function X(e){let t=e.source??"followups",r=()=>typeof e.order=="string"?e.order:String(e.order?.value??"");return new f({pageSize:e.pageSize??20,load:async({page:o,pageSize:n})=>{let c=t==="campaign"?await l("leadbay_campaign_call_sheet",{campaign_id:e.campaignId,page:o,count:n,_triggered_by:e.ask}):await l("leadbay_pull_followups",{page:o,count:n,...e.city?{city:e.city}:{},...r()?{order:r()}:{},_triggered_by:e.ask}),w=c.leads??c.items??[];return{items:w,total:c.total_leads??c.pagination?.total??w.length}}})}function ee(e){return new d({load:()=>l("leadbay_team_activity",{weeks:e.weeks??4,_triggered_by:e.ask})})}var te={VERSION:L,configure:O,styles:R,call:l,field:e=>new g(e),action:e=>new s(e),resource:e=>new d(e),list:e=>new f(e),bindSelect:D,bindValue:z,bindAction:q,campaigns:V,outreach:$,note:G,like:B,dislike:Z,leadStatus:j,setStatus:W,sortOrder:U,leadHistory:Y,leadProfile:K,enrichment:J,callList:X,leadList:Q,teamActivity:ee,EPILOGUE_STATUSES:F,LEAD_STATUSES:m,SORT_ORDERS:y};typeof globalThis<"u"&&(globalThis.LeadbayArtifacts=te);})();';
+var ARTIFACT_USAGE_GUIDE = '# Leadbay Artifact Kit \u2014 headless domain components\n\nYou are building a single-file HTML **artifact** the user runs inside cowork. This\nkit gives you **headless view-models** that own a control\'s whole data lifecycle \u2014\nload/populate from a Leadbay call, hold value/state, poll, validate, and\nencapsulate the API call + business rules. **You own 100% of markup/layout/style.**\nThe library renders nothing. Inline the runtime once as a `<script>`; it exposes\none global `window.LeadbayArtifacts` (call it `lb`). Vanilla, no React, no build.\n\nPass every tool you use as the artifact\'s `mcp_tools` so the host permits it.\n\n## Two layers\n\n**Primitives** (generic):\n- `lb.field({ load, options, value, validate, dependsOn })` \u2014 a value + optionally\n  API-populated options. `.value/.setValue/.options/.loading/.error/.valid/.subscribe`.\n- `lb.action({ tool, args, fields, confirm, onSuccess, onError })` \u2014 a write/submit.\n  `.run()/.loading/.error/.lastResult/.subscribe`.\n- `lb.resource({ load, pollEvery?, until?, autoLoad? })` \u2014 one read that may change:\n  load-on-click or poll-until-`until`. `.data/.loading/.refreshing/.error/.done/.load()/.refresh()/.stop()/.subscribe`.\n- `lb.list({ load, pageSize })` \u2014 paginated rows. `.items/.page/.total/.loading/.loadPage(n)/.next()/.prev()/.hasMore/.subscribe`.\n\n`.error` is `{ message, unavailable } | null`. `subscribe(cb)` fires immediately\nthen on every change \u2014 render your own DOM from it.\n\n**Domain components** (pre-wired \u2014 bake in the tool name, arg shape, and footguns):\n\n| Call | Returns | For |\n|---|---|---|\n| `lb.campaigns(ask)` | field | a campaign `<select>`, options from `leadbay_list_campaigns` |\n| `lb.outreach({leadId, ask, status?, note?})` | action | log a call \u2192 `report_outreach` (verification + `_triggered_by` baked in) |\n| `lb.note({leadId, note})` | action | add a note \u2192 `add_note` |\n| `lb.like(leadId)` / `lb.dislike(leadId)` | action | taste signal |\n| `lb.leadStatus(current?)` | field | a status `<select>` (Wanted/Won/Lost/Unwanted) |\n| `lb.setStatus({leadId or leadIds, status, date?, ask})` | action | write the org CRM status \u2192 `set_lead_status` |\n| `lb.leadHistory(leadId, ask)` | resource (lazy) | notes + activities + engagement \u2192 `account_history` |\n| `lb.leadProfile(leadId, ask)` | resource (lazy) | full lead profile \u2192 `research_lead_by_id` |\n| `lb.sortOrder(current?)` | field | a sort `<select>` mirroring the app\'s TableSort |\n| `lb.leadList({lensId?, order?, ask})` | list | a sortable Discover batch \u2192 `pull_leads` |\n| `lb.callList({source:\'followups\'\\|\'campaign\', campaignId?, city?, ask})` | list | a cold-call list (Monitor or a campaign) |\n| `lb.enrichment({leadIds, titles, ask, pollEvery?})` | resource (polling) | launch + watch contact enrichment |\n| `lb.teamActivity({weeks, ask})` | resource | manager leaderboard + activity trend \u2192 `leadbay_team_activity` |\n\n`lb.EPILOGUE_STATUSES` = the 4 disposition values\n(`STILL_CHASING`, `COULD_NOT_REACH_STILL_TRYING`, `INTEREST_VALIDATED_OR_MEETING_PLANED`, `NOT_INTERESTED_LOST`).\n`lb.LEAD_STATUSES` = the 4 org CRM statuses as `{value,label}` (`WANTED`, `WON`, `LOST`, `UNWANTED`).\n`lb.SORT_ORDERS` = the sort options as `{value,label}`; values are the backend `FIELD:ASC|DESC` enum.\n\n**Sorting is a SERVER concern.** `lb.leadList` and `lb.callList` take an `order`\n(a `lb.sortOrder()` field or a literal) and send it upstream; the backend sorts\nthe whole lens / Monitor and returns the requested page of that. Never re-sort\nrows in the browser \u2014 you would be reordering one page of a larger set, showing\nleads that do not belong at that position. The empty value means "no order\nparam", i.e. the tab\'s own ranking, which is the right default. Changing the\nsort should reset to page 0. Campaign call sheets cannot sort:\n`leadbay_campaign_call_sheet` has no `order` param, and `lb.callList` drops it\nfor that source rather than sending something the tool would reject.\n\n**Two different systems.** Epilogue = how one outreach attempt went (drives\nfollow-up ranking). Lead status = the commercial outcome, org-wide \u2014 the same\nfield the website\'s status selector writes. A won deal is a LEAD STATUS;\n"she didn\'t pick up" is an EPILOGUE. Setting one never sets the other, so when\nthe user reports both in one breath, fire both actions.\n\n**Binding sugar** (optional; binds a view-model to YOUR native element, no style):\n`lb.bindSelect(selectEl, field)` (populates options + value), `lb.bindValue(inputEl, field)`,\n`lb.bindAction(buttonEl, action)`. They set `data-lb-state`\n(`ready|loading|error|success|unavailable`) + `data-lb-error` on your element as\nstyling hooks. For lists/resources, use `.subscribe()` and render yourself.\n\n`ask` is the user\'s request this artifact serves \u2014 it becomes `_triggered_by`.\n\n## The skin (optional) \u2014 `lb.styles()`\n\nCall it once and you get a small `lb-*` stylesheet, so every artifact you build\nshares one visual language instead of re-inventing padding and colours. It is\n**opt-in**: skip it and you get exactly the unstyled HTML you wrote. It injects\nno markup and never touches your `class` attributes.\n\n```js\nlb.styles();   // idempotent \u2014 safe to call per row\n```\n\n| Class | For |\n|---|---|\n| `lb-card` / `lb-card-head` / `lb-title` / `lb-sub` | a lead card + its header |\n| `lb-row` / `lb-stack` / `lb-spacer` | control row / vertical spacing / flex filler that right-aligns what follows |\n| `lb-link-out` | quiet external link (icon inherits currentColor) \u2014 "Open in Leadbay" |\n| `lb-select` / `lb-input` / `lb-btn` | form controls (state-aware, see below) |\n| `lb-msg` (`data-tone="error\\|ok"`) | inline feedback |\n| `lb-chip` (`data-status="WON\\|LOST"`) | a status pill |\n| `lb-table` | leads table |\n| `lb-spinner` | inline busy indicator \u2014 decorative, mark it `aria-hidden` |\n| `lb-vh` | visually-hidden text \u2014 labels heard but not seen |\n\nControls react to the `data-lb-state` the bind helpers already set \u2014 a bound\n`lb-btn` dims while loading, goes green on success, red on error, all with no\nextra CSS from you.\n\nThe palette is the **product design system**, ported from\n`frontend/packages/style/color.css` \u2014 same `--color-gray-1\u20269` ramp, same\nsemantic `--color-{green,red,blue,gold}-{background,foreground}` pairs, same\n`1.5rem` / `0.625rem` radii (concentric: outer = inner + padding) and\n`corner-shape: squircle`, matching the app\'s components.\nAn artifact therefore looks like Leadbay, not like a generic page.\n\nUse the tokens rather than hardcoded colours \u2014 the same rule the style package\nenforces. Re-theme by overriding them; don\'t fight specificity:\n\n```css\n:root { --lb-surface: var(--color-gray-2); --lb-radius: 0.5rem; }\n```\n\nDark mode works two ways: `data-theme="dark"` on `<html>` (the frontend\'s own\nhook) **and** `prefers-color-scheme`, because an artifact renders inside a host\nwhose theme attribute it cannot set. Never hardcode a light background over the\nskin.\n\nThe product face is `Nikkei Maru`; the stack names it first and falls back to\nthe system UI font. Do **not** add an `@font-face` \u2014 artifacts are inline-only\nand a remote font URL will silently fail.\n\n## What every lead card MUST carry\n\nA card is the artifact form of the `pull_leads` table, and it inherits that\ntable\'s rules. A card with a name and a button is not enough: the rep cannot\ntell *why* this lead is on screen. Five lines, in this order.\n\n```html\n<div class="lb-card">\n  <div class="lb-card-head">\n    <span class="lb-title"></span>              <!-- 1. company -->\n    <span class="lb-chips">                     <!-- 2. state -->\n      <span class="lb-chip" data-taste hidden></span>\n      <span class="lb-chip" data-status hidden></span>\n    </span>\n  </div>\n  <div class="lb-facts">                        <!-- one group, tight 4px gap -->\n    <div class="lb-sub"></div>                  <!-- 3. firmographics -->\n    <div class="lb-sub" data-who></div>         <!-- 4. the person -->\n    <div class="lb-sub" data-how></div>         <!-- 5. company channels -->\n  </div>\n  <div class="lb-sub" data-why></div>           <!-- 6. why it fits -->\n  <div class="lb-row"><!-- actions --></div>\n  <div class="lb-msg" role="status" aria-live="polite"></div>\n</div>\n```\n\n1. **Company** \u2014 `name`, linked to `website` (prefix `https://` on a bare host).\n   Never render the numeric `score`; use the `\u25B0\u2756\u25B1` bar if you want the signal.\n\n   Keep `.lb-msg` OUT of `.lb-row`. The result of a write \u2014 "Could not reach the\n   host" \u2014 is the most important thing on the card at that moment; parked between\n   two buttons it reads as a control, and as a wide flex item it forces the\n   trailing link onto a line of its own. Give it its own row after the actions.\n\n   Also give every card an **Open in Leadbay** link to the lead\'s panel in the\n   product. Put it at the **trailing end of the card\'s last action row** \u2014\n   same row as the buttons, pushed right by an `lb-spacer`, not on a line of\n   its own. Style it `lb-link-out`: quiet text plus a plain arrow-up-right,\n   never a filled button. It is an escape hatch, not a call to action.\n\n   Group the controls by what they act on. A flat row of five buttons reads as\n   five peers; the rep cannot see that "Set status" commits the select beside it\n   while Like/Dislike are independent toggles. Wrap each axis in an `lb-group`\n   and mark the commit with `lb-btn-submit`.\n\n   Taste is the one pair worth reducing to icons: thumbs up/down are unambiguous,\n   they repeat on every card, and dropping the words buys the width a narrow chat\n   host needs. Use `lb-btn-icon` \u2014 and note the three attributes it REQUIRES,\n   because with no text the glyph is the whole affordance:\n\n   - `aria-label` naming the lead ("Like Acme Corp"), since the control repeats\n     N times down the list;\n   - `title` so a sighted user who does not know the glyph gets a tooltip;\n   - `aria-pressed` reflecting the current taste \u2014 a toggle must say whether it\n     is on, and `[aria-pressed=true]` is what the skin styles.\n\n   Do **not** reduce "Set status" to an icon: no glyph says "commit the value in\n   the select beside me". Icons work for a fixed, well-known action; they fail\n   for one whose meaning comes from a neighbouring control.\n\n   ```html\n   <div class="lb-row">\n     <span class="lb-group">                  <!-- taste: two toggles -->\n       <button class="lb-btn lb-btn-icon" aria-label="Like Acme Corp"\n               title="Like" aria-pressed="false">\n         <svg viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2"\n              stroke-linecap="round" stroke-linejoin="round" aria-hidden="true">\n           <path d="M19 14c1.49-1.46 3-3.21 3-5.5A5.5 5.5 0 0 0 16.5 3c-1.76 0-3 .5-4.5 2-1.5-1.5-2.74-2-4.5-2A5.5 5.5 0 0 0 2 8.5c0 2.3 1.5 4.05 3 5.5l7 7Z"/>\n         </svg>\n       </button>\n       <button class="lb-btn lb-btn-icon" data-taste="disliked"\n               aria-label="Dislike Acme Corp" title="Dislike" aria-pressed="false">\n         <svg viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2"\n              stroke-linecap="round" stroke-linejoin="round" aria-hidden="true">\n           <path d="M17 14V2"/>\n           <path d="M9 18.12 10 14H4.17a2 2 0 0 1-1.92-2.56l2.33-8A2 2 0 0 1 6.5 2H20a2 2 0 0 1 2 2v8a2 2 0 0 1-2 2h-2.76a2 2 0 0 0-1.79 1.11L12 22a3.13 3.13 0 0 1-3-3.88Z"/>\n         </svg>\n       </button>\n     </span>\n     <select class="lb-select" aria-label="Lead status for Acme Corp"></select>\n     <span class="lb-spacer"></span>          <!-- pushes the link to the end -->\n     <a class="lb-link-out" data-k="open" target="_blank" rel="noopener">\n       Open in Leadbay\n       <svg viewBox="0 0 24 24" fill="none" stroke="currentColor" stroke-width="2"\n            stroke-linecap="round" stroke-linejoin="round" aria-hidden="true">\n         <line x1="7" y1="17" x2="17" y2="7"/><polyline points="7 7 17 7 17 17"/>\n       </svg>\n     </a>\n   </div>\n   ```\n\n   Keep the arrow a bare diagonal stroke \u2014 the text already says where the link\n   goes, so the glyph only has to mark "leaves this page". Mark the `<svg>`\n   `aria-hidden="true"`: it is decorative, and the link text is the accessible\n   name.\n   **Pick the view the lead actually lives in** \u2014 the URL is\n   `/app/<view>?lead=<uuid>`, and the three views are `discover`, `monitor`,\n   `campaign`. Landing a Monitor lead on Discover drops the rep into a list\n   that does not contain it:\n\n   ```js\n   function leadUrl(lead, campaignId) {\n     const id = encodeURIComponent(lead.id);\n     // A campaign card carries TWO params \u2014 the campaign selects the list, the\n     // lead opens the panel inside it. Campaign wins even when in_monitor is\n     // also true, because that is the list the rep is looking at.\n     if (campaignId) {\n       return `https://leadbay.app/app/campaign?campaign=${encodeURIComponent(campaignId)}&lead=${id}`;\n     }\n     const view = lead.in_monitor ? "monitor" : "discover";\n     return `https://leadbay.app/app/${view}?lead=${id}`;\n   }\n   openEl.href = leadUrl(lead, campaignId);\n   ```\n\n   `in_monitor` / `in_discover` are booleans on the `pull_followups` payload \u2014\n   every follow-up carries `in_monitor: true`, so a call sheet must link to\n   `monitor`. `pull_leads` omits both flags entirely; its leads are the Discover\n   batch by definition, so `discover` is the default. A campaign card\n   (`lb.callList({source:"campaign", campaignId})`) needs `?campaign=<id>&lead=<id>`\n   \u2014 the param names are `CAMPAIGN_QUERY_PARAM` and `LEAD_QUERY_PARAM`, and the\n   app\'s own `useLeadPanel` preserves whatever params are already set, so the\n   two coexist by design. Omitting `campaign=` opens an empty campaign view.\n\n   Inline the glyph as SVG rather than an emoji or `\u2197` \u2014 it inherits\n   `currentColor` and scales with the text, so it stays legible in both themes.\n   `?lead=<uuid>` is the real deep-link (`LEAD_QUERY_PARAM` in the web app, read\n   on load; the panel is an overlay, so the view choice only decides what sits\n   behind it). This is the ONE place a card may use `lead.id`: as a link target,\n   never as visible text.\n2. **State chips** \u2014 taste (`data-taste`) and CRM status (`data-status`) are\n   INDEPENDENT axes; render both, hide the empty one. Never collapse to one chip.\n3. **Firmographics** \u2014 sector of activity first, then city, then size, then the\n   contact. `sector_id` is a RAW ID (`"5136"`), not a label: resolve it via\n   `leadbay_list_sectors` (1346 rows \u2014 fetch once, cache, never inline the lot)\n   or omit it. Never print the raw id.\n\n   **Always show whether the lead is reachable \u2014 and never merge the person\n   with the company\'s switchboard.** These are two separate lines:\n\n   ```js\n   // WHO \u2014 recommended_contact. Name, and job_title ONLY when present; on list\n   // payloads it is usually null, and inventing one is worse than omitting it.\n   const rc = lead.recommended_contact;\n   const who = rc ? [rc.first_name, rc.last_name].filter(Boolean).join(" ") : null;\n   const whoLine = who ? who + (rc.job_title ? " \xB7 " + rc.job_title : "") : "No contact yet \u2014 enrich to find one";\n\n   // HOW \u2014 company-level channels. `phone_numbers` and `email` belong to the\n   // COMPANY, not to `recommended_contact`. Rendering "Jean \xB7 \u260E 0123\u2026" claims a\n   // direct line that does not exist; it is the switchboard.\n   // The API returns the literal STRING "null" for a missing value \u2014 in\n   // phone_numbers as well as email (a real lead ships phone_numbers:["null"]).\n   // Guard BOTH or the card prints "\u260E null" as if it were a number.\n   const real = (v) => (v && v !== "null" ? v : null);\n   const phone = real((lead.phone_numbers || [])[0]);\n   const email = real(lead.email);\n   const howLine = [phone && "\u260E " + phone, email && "\u2709 " + email].filter(Boolean)\n     .join(" \xB7 ") || "No phone or email \u2014 enrich to look for them";\n   ```\n\n   ```html\n   <div class="lb-sub">Sector \xB7 City \xB7 Size</div>\n   <div class="lb-sub"><span aria-hidden="true">\u{1F464}</span> Jean-Fran\xE7ois Froemer \xB7 G\xE9rant</div>\n   <div class="lb-sub"><span class="lb-vh">Company switchboard: </span><span aria-hidden="true">\u{1F3E2} \u260E</span> 01 23 45 67 89 (company line)</div>\n   ```\n\n   Label the channel line as the **company\'s**, so a rep reading fast cannot\n   mistake it for a direct line. A per-contact email or phone exists only after\n   enrichment \u2014 `research_lead_by_id` exposes it as `contacts.reachable[]`, and\n   `_meta.has_reachable_contact` is the authoritative flag. The list payloads\n   carry neither, so a card built from `pull_leads` / `pull_followups` can only\n   ever show company channels. Say "enrich to reveal" rather than implying the\n   contact is callable.\n\n   Two things that look like reachability and are not: a `linkedin_page` alone\n   (the rep cannot message a URL without leaving the artifact \u2014 same rule\n   `research_lead_by_id` applies), and `contacts_count > 0` (it counts known\n   people, not people you can contact; a lead can show 2518 contacts and zero\n   channels). `pull_followups` carries `has_phone` as a ready-made boolean;\n   `pull_leads` omits it, so derive from `phone_numbers` there.\n4. **Why it fits** \u2014 one sentence, \u226420 words. Walk this chain and stop at the\n   first hit:\n\n   1. `short_description`\n   2. `description` (longer; only on `research_lead_by_id` /\n      `research_lead_by_name_fuzzy` \u2014 the trim payloads omit it)\n   3. top 2 `tags[].display_name`\n   4. `qualification_summary.best_response_excerpt`, trimmed to one sentence\n   5. `keywords`, first 3, joined with ` \xB7 `\n   6. the resolved sector label \u2014 better than nothing, and if step 3 already\n      printed the sector on the firmographics line, skip to step 7\n   7. the literal *"No description yet \u2014 qualify to add one"* ("qualify" is the\n      product\'s own verb; "run qualification to generate one" is a nominalisation)\n\n   Never leave this line blank: a silent gap reads as a rendering bug, whereas\n   the fallback tells the rep the data is missing and what fixes it.\n\n   **The two list payloads are complementary, so the chain must span both.**\n   `pull_leads` returns `short_description` on every lead but no `sector_id`;\n   `pull_followups` returns `sector_id` but no `short_description` at all. A\n   card fed by one will fall through to a different step than the same card fed\n   by the other \u2014 that is expected, not a bug. Never call\n   `research_lead_by_id` per row just to fill this line: it is one request per\n   lead. Fetch it lazily when the rep expands a card.\n\n**Never show** on a card: `id`, `sector_id`, `location.pos`, `location.country`\n(unless city and state are both missing), `is_hq`, `*_in_progress`,\n`highlighted_fields`, `custom_fields`, `stale_at`, `deal_insights`,\n`need_attention*`, any count that is 0, any value that is the string `"null"`.\n\n**Minimum actions.** A card that only displays is a table row that costs more \u2014\nif you are not wiring an action, render the markdown table instead. Wire at\nleast one write, and prefer the set the rep actually needs:\n\n| Card is for | Wire |\n|---|---|\n| triage a discovery batch | `lb.like` / `lb.dislike` + `lb.setStatus` |\n| working a call list | `lb.outreach` (gated on a note) + `lb.leadHistory` |\n| pipeline review | `lb.setStatus` + `lb.note` |\n\nAlways render the `.error` branch of every view-model \u2014 a control that cannot\nreach the host must say so, not sit silent.\n\n   **Accessibility is the markup\'s job, not the skin\'s.** `data-lb-state` is a\n   STYLING hook; it sets no ARIA. A card renders N times in a list, so every\n   repeated control needs a name that says *which* lead it acts on:\n\n   ```html\n   <select class="lb-select" aria-label="Lead status for Acme Corp"></select>\n   <button class="lb-btn" aria-label="Like Acme Corp">Like</button>\n   <a class="lb-link-out" aria-label="Open Acme Corp in Leadbay" \u2026>\n   <span class="lb-msg" role="status" aria-live="polite"></span>\n   ```\n\n   Without the `role="status"` node the rep hears nothing when a write fails \u2014\n   `bindAction` puts the message in a `data-lb-error` attribute that nothing\n   renders. Mark the `\u25B0\u2756\u25B1` bar `aria-hidden="true"` and follow it with\n   `<span class="lb-vh">Fit: strong</span>`; the glyphs otherwise read aloud as\n   "black parallelogram black parallelogram\u2026" before the company name. Use\n   `lb-vh` for any label that should be heard but not seen.\n\n## Recipe: cold-call sheet (one row per lead)\n\n```js\nconst lb = window.LeadbayArtifacts; lb.configure();\nconst ASK = "<the user\'s request>";\n\nconst list = lb.callList({ source: "campaign", campaignId: CID, ask: ASK });\nlist.subscribe((l) => renderRows(l.items, l.loading));   // your render\n\n// per lead row (call when you build a row):\nfunction wireRow(lead, els) {\n  const status = lb.field({ value: "STILL_CHASING" });   // static-enum <select>\n  const note   = lb.field({ validate: (v) => (v && v.trim() ? null : "Add a note") });\n  lb.bindValue(els.status, status);\n  lb.bindValue(els.note, note);\n  lb.bindAction(els.log,  lb.outreach({ leadId: lead.id, ask: ASK, status, note }));\n  lb.bindAction(els.like, lb.like(lead.id));\n\n  const history = lb.leadHistory(lead.id, ASK);          // lazy\n  history.subscribe((h) => renderHistory(els.history, h));\n  els.expand.onclick = () => history.load();             // load on click\n}\n```\n\n## Recipe: lead-status dropdown (Wanted / Won / Lost)\n\nThe org-wide CRM status, as a `<select>` + Apply button. You write the markup;\n`lb.leadStatus` fills the options and holds the value, `lb.setStatus` does the write.\n\n```html\n<div class="lb-card">\n  <div class="lb-card-head">\n    <span class="lb-title">Acme Corp</span>\n    <span class="lb-chips">\n      <span id="taste" class="lb-chip" data-taste="liked">Liked</span>\n      <span id="crm"   class="lb-chip" data-status="WANTED">Wanted</span>\n    </span>\n  </div>\n  <div class="lb-row">\n    <select id="st" class="lb-select"></select>\n    <button id="go" class="lb-btn">Apply</button>\n    <span id="msg" class="lb-msg"></span>\n  </div>\n</div>\n```\n\n**Two badges, never one.** Taste (`liked`/`disliked`, from `lb.like`/`lb.dislike`)\nand CRM status (`WANTED`/`WON`/`LOST`/`UNWANTED`, from `lb.setStatus`) are\nindependent axes \u2014 a lead can be liked *and* lost. Collapsing them into a single\nchip destroys information: the rep can no longer see that a lead they liked went\nnowhere. Render `data-taste` and `data-status` as separate chips inside\n`lb-chips`, and hide the one that has no value rather than reusing it.\n\n```js\nlb.styles();                                          // once per artifact \u2014 see below\n\nconst status = lb.leadStatus(lead.org_lead_status);   // seed with the CURRENT value\nconst save   = lb.setStatus({ leadId: lead.id, status, ask: ASK });\n\nlb.bindSelect(document.getElementById("st"), status); // populates the 4 options\nlb.bindAction(document.getElementById("go"), save);   // click \u2192 write\n\nsave.subscribe((a) => {                               // render your own feedback\n  msg.textContent = a.loading ? "Saving\u2026"\n    : a.error ? a.error.message                       // includes partial failures\n    : a.lastResult ? `Set to ${a.lastResult.status}` : "";\n  msg.dataset.tone = a.error ? "error" : a.lastResult ? "ok" : "";\n});\n```\n\nLoading / success / error styling comes free: `bindAction` and `bindSelect` set\n`data-lb-state` (`ready|loading|error|success|unavailable`) and the skin already\ntargets those attributes. No extra wiring.\n\n**Save on change is the default for status.** The rep picks a value and it\nwrites \u2014 one interaction, no second button, matching the web app\'s own status\nselector. Drop `bindAction` and run the action from the change event:\n\n```js\nconst sel = document.getElementById("st");\nsel.addEventListener("change", () => save.run());\nsave.subscribe((a) => {                       // the select IS the feedback surface\n  sel.setAttribute("data-lb-state",\n    a.loading ? "loading" : a.error ? "error" : a.lastResult ? "success" : "ready");\n  msg.textContent = a.loading ? "Saving\u2026" : a.error ? a.error.message\n    : a.lastResult ? "Saved" : "";\n  msg.dataset.tone = a.error ? "error" : a.lastResult ? "ok" : "";\n});\n```\n\nWithout a submit button the select becomes the only affordance, so it MUST show\nthe write: mirror `data-lb-state` onto it (the skin already styles loading /\nsuccess / error on `.lb-select`) and put the outcome in the `role="status"`\nline. A silent select leaves the rep unsure whether anything happened.\n\nKeep a submit button ONLY where a mis-click is expensive and the value is not\nself-evident \u2014 a bulk apply across checked rows, for instance, which already\ntakes a `confirm`. For one lead, the extra step buys nothing: the value is\nvisible in the select, and the rep can simply pick again.\n\n**Bulk apply** across checked rows \u2014 pass `leadIds` and a `confirm`, since one\nclick rewrites a field every rep in the org sees:\n\n```js\nconst bulk = lb.setStatus({\n  leadIds: () => checkedIds,        // \u2190 read at run() time, not at build time\n  status, ask: ASK,\n  confirm: "Set this status on every selected lead?",\n});\n```\n\n`leadIds` is read when the action runs, so a live selection works \u2014 but pass the\narray itself if your selection is fixed. A partial write (some leads rejected)\nsurfaces as `.error`, never as a green button: `setStatus` checks the `failed[]`\nthe tool returns.\n\nThe backend stamps the status date as "now" on every write, which is what a rep\nclicking a dropdown means. Don\'t add a date picker unless the user asks to\nbackdate \u2014 then pass an optional `date` field holding `YYYY-MM-DD`:\n`lb.setStatus({ leadId, status, date, ask })`.\n\n## Recipe: manager dashboard\n\n```js\nconst team = lb.teamActivity({ weeks: 4, ask: ASK });\nteam.subscribe((t) => {\n  if (t.loading) showSpinner();\n  if (t.data) {\n    renderLeaderboard(t.data.reps);   // sorted by total_activities; cols: name, notes, meetings_or_interest, lost\u2026\n    renderTrendChart(t.data.trend);   // [{date,count}] \u2192 Chart.js (allowed from CDN)\n  }\n});\nrefreshBtn.onclick = () => team.refresh();\n```\n\n## Recipe: live enrichment\n\n```js\nconst job = lb.enrichment({ leadIds: [LEAD], titles: ["CEO", "VP Sales"], ask: ASK });\njob.subscribe((j) => {\n  const p = j.data && j.data.overall_progress;            // {done,total,done_ratio}\n  renderBar(p);\n  if (j.done) renderContacts(j.data.leads);               // enriched contacts\n});\nrefreshBtn.onclick = () => job.refresh();\n```\n\n## Write-call rules\n\nThe domain factories handle these for you. If you hand-roll an action:\n`leadbay_report_outreach` args MUST include `verification:{source:"user_confirmed", ref}`\nAND `_triggered_by`; `leadbay_add_leads_to_campaign` needs `_triggered_by`;\n`add_note`/`like_lead`/`dislike_lead` take only their own args. `epilogue_status` is\none of `lb.EPILOGUE_STATUSES`. Snoozing (pushback) is advanced-gated \u2014 not\ncallable from a default artifact. Org lead status IS on the default surface:\nuse `lb.setStatus`, which owns the arg shape AND the partial-write check \u2014\n`leadbay_set_lead_status` writes each lead separately, so it can resolve 200\nwith a non-empty `failed[]`. Hand-rolling that action will report a green\nbutton over a write that never landed.\n\n## Degradation + live updates\n\nIf the host bridge is absent, a view-model\'s `.error` is set with `.error.unavailable\n=== true` (bind helpers set `data-lb-state="unavailable"`) \u2014 nothing throws. Every\ncall also has a **30s timeout** (configurable via `lb.configure({ timeoutMs })`): a\nhost call that never settles becomes `.error` with `code:"timeout"`, so a control is\nnever stuck loading forever \u2014 always render the `.error` branch so the user can retry.\nAuto-poll (`pollEvery`) depends on the cowork host serving FRESH reads; `.refresh()`\nis the guaranteed manual path \u2014 always wire a Refresh control for polling resources.';
 
 // ../core/dist/tools/artifact-kit.js
 var artifactKit = {
@@ -27027,7 +27273,7 @@ function buildAcknowledgeUpdateTool(opts) {
 var ENRICHMENT_TERMINAL = `A settled-empty enrichment is TERMINAL \u2014 do not re-attempt it on a later run. On any contact record, \`enrichment.done: true\` together with \`enrichment.credits_used: 0\` means the reveal completed and the provider returned nothing. Roughly 29% of all enrichments land there, so treat it as an ordinary outcome, not an anomaly. The answer will not change tomorrow: a scheduled task that keeps calling leadbay_enrich_titles on the same contacts every run is buying nothing and hiding its own lack of progress. Read the two fields TOGETHER \u2014 \`credits_used: 0\` is also what an IN-FLIGHT reservation reports (\`done: false\`), and an ABSENT \`credits_used\` means the cost is unknown, not zero; neither is a verdict on its own. \`enrichment: null\` is a different state again \u2014 that contact was never requested and IS enrichable. The one retryable exception is a launch that errored in this same session: its reservation settles as a zero-credit failure, so retry it once, then treat it as terminal. When you report to the user, say what actually happened \u2014 "we looked for a contact at <company> and there isn't one we can reach" \u2014 instead of presenting it as still pending or silently retrying it.`;
 var FRICTION = `Problem reports: when the user asks you to report a Leadbay problem ("report this", "tell the team this didn't work"), call leadbay_report_friction with {category, message (the user's own words), tool_called?, severity?}. If they stated the problem in the same breath as the request, those words ARE the message \u2014 send it in that turn rather than asking them to confirm wording they just gave you, and never stall on optional fields (omit what you don't know). If you notice a problem worth reporting but the user hasn't asked, OFFER once \u2014 "Want me to report this to the Leadbay team?" \u2014 and call it only if they agree. Never call it unprompted. Always tell the user the outcome the tool returns: if \`reported\` is false the report was NOT delivered and you must say so rather than implying it was sent. Frustration alone is not a reason to call it: keep solving their ask.`;
 var MENTAL_MODEL = `How Leadbay works (mental model): Leadbay is a sales inbox, not a queryable database. Each day the user logs back in, a fresh batch of leads is delivered. Batch size is paced by how many leads the user has actually acted on recently \u2014 some workflows produce a big stream of smaller prospects, others a narrow stream of bigger ones. Pulling more won't produce more; the user acting on leads (outreach, skips, saves) does.`;
-var QUOTA_REFRESH = `Show the refreshed quota AFTER a paid action has actually COMPLETED: when leadbay_bulk_enrich_status reports the job done \u2014 all_done, OR a plateau you've decided is terminal (overall_progress.done stopped climbing across spaced polls, so some contacts are unresolvable and all_done stays false) \u2014 OR a top-up the user confirmed landed, call leadbay_account_status once and render the refreshed quota \u2014 the per-window %/$ gauge (Daily/Weekly/Monthly) it returns \u2014 so the user sees where they now stand. Wait for genuine completion: leadbay_enrich_contacts only LAUNCHES an async reveal (it returns a hint to check back in ~60s), so do NOT refresh quota right after it \u2014 the usage isn't reflected yet. For that single-contact flow, refresh only once a re-read of the lead's contacts (leadbay_research_lead_by_id; leadbay_get_contacts where exposed) shows the REQUESTED channel actually landed \u2014 the requested email and/or phone_number present \u2014 NOT enrichment.done alone (that flag is already true for a contact enriched on the other channel earlier, so a phone reveal could otherwise trigger the refresh before phone_number arrives). This is the canonical quota surface; do NOT hand-roll a 'credits' line in its place. Skip it only when account_status reports unlimited_credits, quota_error, or a null quota (nothing to show), or when billing is genuinely unavailable. Do it ONCE per completed action \u2014 not after every poll while work is still in progress.`;
+var QUOTA_REFRESH = `Show the refreshed quota AFTER a quota-using action only when it matters: the user asks where they now stand, the action stopped on an exhausted window, or a top-up the user confirmed landed. Otherwise report the result and say nothing about quota. When you do show it, wait for genuine completion \u2014 leadbay_bulk_enrich_status reports the job done (all_done, OR a plateau you've decided is terminal: overall_progress.done stopped climbing across spaced polls, so some contacts are unresolvable and all_done stays false) \u2014 then call leadbay_account_status once and render the per-window %/$ gauge (Daily/Weekly/Monthly) it returns. leadbay_enrich_contacts only LAUNCHES an async reveal (it returns a hint to check back in ~60s), so do NOT refresh quota right after it \u2014 the usage isn't reflected yet. For that single-contact flow, refresh only once a re-read of the lead's contacts (leadbay_research_lead_by_id; leadbay_get_contacts where exposed) shows the REQUESTED channel actually landed \u2014 the requested email and/or phone_number present \u2014 NOT enrichment.done alone (that flag is already true for a contact enriched on the other channel earlier, so a phone reveal could otherwise trigger the refresh before phone_number arrives). This is the canonical quota surface; do NOT hand-roll a 'credits' line in its place. Skip it when account_status reports unlimited_credits, quota_error, or a null quota (nothing to show), or when billing is genuinely unavailable. Do it ONCE per completed action \u2014 not after every poll while work is still in progress.`;
 var QUOTA_TOPUP = `Quota & top-ups: when a tool returns QUOTA_EXCEEDED / 429, the user has TWO options \u2014 wait for the window reset (daily / weekly / monthly resets shown in leadbay_account_status), OR top up AI credits (top-ups clear the throttle IMMEDIATELY \u2014 they are not subject to the same window). Always offer BOTH options; default-recommending 'wait until tomorrow' is wrong when a 30-second top-up unblocks the same call. If the host exposes leadbay_create_topup_link, OFFER it on every quota wall: 'Want me to generate a top-up link?' \u2014 when the user says yes, call leadbay_create_topup_link and surface the returned Stripe URL as a clickable link for the user to open in their browser. (Sibling leadbay_open_billing_portal is for ongoing subscription changes, not one-shot top-ups.) AFTER the user has topped up: do NOT keep refusing operations. A top-up invalidates every prior 429 and every stale 'you're at your quota' snapshot. The moment the user signals they topped up / bought credits / added credits \u2014 even WITHOUT re-calling account_status \u2014 treat the previous quota state as void and RETRY the originally failed call. (Best practice: re-call leadbay_account_status to surface the fresh state to the user, then retry; but the retry itself does NOT require a successful account_status check first. If the retry hits the wall again, THEN you have evidence the top-up didn't land; only then re-offer top-up / wait.) The agent's job after a top-up is to RESUME the workflow the user was on, not gate-keep.`;
 var TRANSIENT_401 = `Transient 401s (DO NOT narrate as an auth problem): Leadbay bearer tokens do NOT expire on a timer, and the client already auto-retried this call once before surfacing it. So a lone AUTH_EXPIRED / 401 from a tool is almost always a brief Leadbay-side hiccup, NOT a sign the user must sign in again. When you see ONE, just silently retry the operation once more and carry on \u2014 do NOT tell the user to log in again, reconnect the connector, reauthorize, or 're-load Leadbay', and do NOT surface the bare '401' to them at all. Only if calls keep failing across several attempts should you mention it \u2014 and then frame it as a possible Leadbay-side issue or a possible logout (you cannot tell which from here), and offer to report it to the team. Never turn a one-off 401 into a 'your connection needs re-authenticating' message; that is the wrong read on a connection that is actually working.`;
 var TRIGGERED_BY = `Trigger provenance (MANDATORY): every Leadbay composite-tool call MUST carry a non-empty \`_triggered_by\` argument \u2014 the verbatim slice of the user's most recent message that this call is acting upon. Quote literally; do NOT paraphrase, summarize, or substitute a one-word label like "leads" or "request" (those are rejected). If you are acting WITHOUT a fresh user message (a memory recall, a scheduled run, a self-initiated retry), pass the actual instruction you are acting on \u2014 the recalled directive, the schedule's intent, or the original request being retried \u2014 so the value is always a real, auditable trace. Strip any secrets the user pasted (API keys, passwords, card numbers, full home addresses) \u2014 replace with [REDACTED]. A composite call missing or blanking this field is rejected with LAST_PROMPT_REQUIRED; just re-call with the field set. This is a protocol requirement on EVERY composite invocation (not just the first), independent of any telemetry setting.`;
@@ -27166,7 +27412,7 @@ function buildProtocolPrimitivesParagraph(has) {
     }
     if (deliveryRunners.length > 0) {
       clauses.push(
-        "On " + deliveryRunners.map((n) => `leadbay_${n}`).join(", ") + " the job is BACKEND-owned and likewise keeps running. Any work already paid for still completes; poll `leadbay_lead_job_status` with the `job_id` later to collect it."
+        "On " + deliveryRunners.map((n) => `leadbay_${n}`).join(", ") + " the job is BACKEND-owned and likewise keeps running. Any work already launched still completes; poll `leadbay_lead_job_status` with the `job_id` later to collect it."
       );
     }
     parts.push(
@@ -28126,7 +28372,7 @@ function parseWriteEnv(env = process.env) {
 }
 
 // src/http-server.ts
-var VERSION = true ? "0.38.0" : "0.0.0-dev";
+var VERSION = true ? "0.38.2" : "0.0.0-dev";
 var PORT = Number(process.env.PORT ?? 8080);
 var HOST = process.env.HOST ?? "0.0.0.0";
 var logger = {