@sellable/mcp 0.1.144 → 0.1.146

package/skills/load-voice/SKILL.md

@@ -0,0 +1,129 @@
+---
+name: load-voice
+description: Load Sellable voice/company memory before answering questions, writing posts, applications, replies, or other copy in the user's voice.
+visibility: public
+allowed-tools:
+  - Read
+  - Glob
+  - Grep
+---
+
+# Load Sellable Voice
+
+<role>
+You are the Sellable voice loader. Your job is to read the user's durable
+Sellable voice/company memory into the current thread before drafting,
+answering, rewriting, reviewing, or calibrating copy in the user's voice.
+
+This is a read-only usage workflow. Do not interview by default, do not write
+memory files, and do not update `.sellable/**` unless the user explicitly asks
+to switch into `$sellable:interview`.
+</role>
+
+<default_use_cases>
+Use this workflow when the user asks to:
+
+- answer questions on behalf of them
+- write or rewrite a LinkedIn post
+- answer an application, investor, customer, or sales question
+- draft a sales reply, objection response, cold note, or website line
+- review whether copy sounds like them
+- "use my voice", "load my voice", "answer like me", or "write this as me"
+  </default_use_cases>
+
+<load_order>
+Read only relevant files. Start with the active `.sellable/configs` memory
+layer, not random repo docs.
+
+If `.sellable/configs` is not visible from the current working directory, use
+`Glob` to find `**/.sellable/configs/core/my-company.md`, excluding
+`node_modules`, `.next`, `dist`, and `.git`, then use the nearest parent as the
+Sellable memory root.
+
+## Core Voice And Judgment
+
+Read these when present:
+
+- `.sellable/configs/core/about-me.md`
+- `.sellable/configs/core/anti-ai-writing-style.md`
+- `.sellable/configs/core/decision-rules.md`
+- `.sellable/configs/core/context-modes.md`
+
+## Company Truth And Proof
+
+Read these when present:
+
+- `.sellable/configs/core/my-company.md`
+- `.sellable/configs/core/answer-bank.md`
+- `.sellable/configs/core/proof-ledger.md`
+- `.sellable/configs/core/story-bank.md`
+- `.sellable/configs/core/wins-ledger.md`
+
+## Writing-Specific Guidance
+
+Read these when present and relevant to the task:
+
+- `.sellable/configs/writing/styleguide-core.md`
+- `.sellable/configs/writing/comments.md`
+
+## Transcript Or Topic Context
+
+Do not bulk-read raw transcripts. If the user asks about a specific topic,
+story, sales call, customer, market view, or correction, first read:
+
+- `.sellable/configs/core/transcripts/INDEX.md`
+- the most relevant file under `.sellable/configs/core/transcripts/topics/`
+
+Only read raw archives under `.sellable/interviews/**` when the index/topic file
+points to one and the current task needs that source detail.
+</load_order>
+
+<operating_rules>
+
+- Keep the loader lightweight. The first pass should usually be 6-10 files, not
+  every file under `.sellable`.
+- Apply the loaded profile silently after a short confirmation. Do not paste a
+  long summary of the memory unless the user asks for one.
+- If the user already gave a concrete writing task, load the memory and then do
+  the task. Do not stop at "I loaded the files."
+- If the user only asked to load voice, say what you loaded, name the strongest
+  active rules in 3-6 bullets, then ask what they want drafted, answered, or
+  reviewed.
+- If exact proof, numbers, customer names, or timelines are not confirmed in
+  the loaded memory, preserve that uncertainty. Do not launder uncertain proof
+  into clean claims.
+- Use the current `context-modes.md` behavior for the requested surface:
+  LinkedIn post, sales reply, website, investor/application answer, outbound, or
+  internal strategy.
+- When answering on behalf of the user, write in first person unless the user
+  asks otherwise.
+- If a question cannot be answered from memory, make the smallest honest
+  assumption or ask one specific follow-up. Do not invent private history,
+  metrics, customer facts, or beliefs.
+- If the user says the output is wrong, ask what feels off and accept messy
+  dictated notes. Use that correction in the current thread. If they want the
+  correction saved durably, route them to `$sellable:interview`.
+  </operating_rules>
+
+<response_shape>
+After loading memory, use this shape when there is no concrete draft task yet:
+
+```text
+I loaded your Sellable voice/company memory from {n} files. I’ll apply it
+silently unless you override it.
+
+The active rules I’m carrying forward:
+- ...
+
+What should I answer, write, or review?
+```
+
+When there is already a concrete task, keep the confirmation to one sentence,
+then produce the answer or draft.
+</response_shape>
+
+<quality_bar>
+The result should feel like the user has a standing voice file in the thread:
+specific, current, proof-safe, and ready to answer on their behalf without
+forcing them to re-explain the same voice/company context every time.
+</quality_bar>

package/skills/providers/prospeo.md

@@ -62,7 +62,8 @@ search_prospeo({
 - Pass `searchId` on subsequent pages to paginate.
 - Use `import_leads` with `provider: \"prospeo\"` and the `searchId` to create a lead list and start import.
 - **IMPORTANT:** If `import_leads` returns `needsModeSelection: true`, use `AskUserQuestion` to ask "add to existing leads or replace?" Do NOT assume.
-- After the list finishes and the user confirms it looks good, call `confirm_lead_list` with the `jobId` and `targetLeadCount` from `import_leads` to import into the campaign table.
+- Default source target is 300+ good-fit leads, capped at 2,500 source candidates for now.
+- After the list finishes and the user confirms it looks good, call `confirm_lead_list` with the `jobId` and the review-batch `targetLeadCount` from the active campaign defaults to import only the review batch into the campaign table.
 - `import_leads` owns the watched move to `confirm-lead-list` after a lead list/job exists.
 - `confirm_lead_list` owns the watched move to `filter-choice` after the bounded review batch exists.
 - Post-confirm readback order is required:

package/skills/providers/sales-nav.md

@@ -349,9 +349,9 @@ Step 4 - Final Search + Confirm:
 What would you like?"
 
 User: "Save it"
-→ import_leads({campaignOfferId: "cmp_xxx", searchId: "search_xxx", targetLeadCount: 100})
+→ import_leads({campaignOfferId: "cmp_xxx", searchId: "search_xxx", targetLeadCount: 300})
 → (wait for lead list to finish, ask user to confirm)
-→ confirm_lead_list({campaignOfferId: "cmp_xxx", jobId: "<jobId from import_leads>", targetLeadCount: 100})
+→ confirm_lead_list({campaignOfferId: "cmp_xxx", jobId: "<jobId from import_leads>", targetLeadCount: 25})
 → wait_for_campaign_table_ready({campaignId: "cmp_xxx"})
 → get_campaign_context({campaignId: "cmp_xxx", refresh: true}) + get_rows_minimal(...)
 </iteration_example>
@@ -382,6 +382,8 @@ User: "Save it"
 <limits>
 - 25 results per page
 - Maximum 100 pages (2,500 leads)
+- Default source target: 300+ good-fit leads, capped at 2,500 source
+  candidates for now
 - Maximum 5 search calls per session
 - If user needs more than 2,500: explain limit, suggest splitting by region
 </limits>

package/skills/providers/signal-discovery.md

@@ -98,11 +98,12 @@ You must estimate:
   - expected good-fit prospects per right-content post after dedupe/cleanup
 - `postsNeededForTarget`
   - number of right-content posts needed to reach the target good-fit lead
-    count, defaulting to 300 unless the campaign says otherwise
+    count, defaulting to 150 for Signal Discovery unless the campaign says
+    otherwise
 - `requiredEngagersToScrape`
-  - `ceil(targetGoodFitLeads / sampledFitRate)`; for example 20 good-fit
-    prospects per 100 engagers means a 300-good-fit source target needs about
-    1,500 engagers scraped
+  - `ceil(targetGoodFitLeads / sampledFitRate)`; for example 15 good-fit
+    prospects per 100 engagers means a 150-good-fit source target needs about
+    1,000 engagers scraped
 - `projectedRange`
   - the conservative range implied by the observed sample
 
@@ -126,7 +127,7 @@ Use conservative logic:
    - stale posts
    - adjacent communities that are not actual buyers
 4. Explain the pass-through briefly and explicitly:
-   - preferred: "`sampledCount`: 40, `passCount`: 9, `passRate`: ~22%, `goodFitPer100Engagers`: ~22 before cleanup / ~13-16 after cleanup, `requiredEngagersToScrape`: ~1,875-2,300 for a 300-good-fit target after cleanup, `avgReachableEngagersPerPost`: 240, `goodFitProspectsPerPost`: ~31-38, `postsNeededForTarget`: ~8-10, `recentStrongPostCount`: 15-25, `freshEnoughPostCount`: 8-12, `projectedRange`: 300-380 from selected posts"
+   - preferred: "`sampledCount`: 40, `passCount`: 9, `passRate`: ~22%, `goodFitPer100Engagers`: ~22 before cleanup / ~13-16 after cleanup, `requiredEngagersToScrape`: ~940-1,150 for a 150-good-fit target after cleanup, `avgReachableEngagersPerPost`: 240, `goodFitProspectsPerPost`: ~31-38, `postsNeededForTarget`: ~4-5, `recentStrongPostCount`: 15-25, `freshEnoughPostCount`: 8-12, `projectedRange`: 150-190 from selected posts"
    - fallback: "I could not fetch engagers, so this is inferred from post and author quality only"
 5. If the evidence is too weak, say so and return a low-confidence estimate instead of pretending precision.
 6. Do not manually enumerate dozens of engagers in the final answer. Summarize the sample by `sampledCount`, `passCount`, `passRate`, `projectedRange`, and 3-6 representative examples.
@@ -378,7 +379,8 @@ For `create-campaign-v2` source approval, do not treat the default
 `selectionTarget` of 3 posts as enough by itself. Before the final source
 recommendation, estimate source capacity from real sample math:
 
-- source target good-fit leads (default 300 unless the campaign says otherwise)
+- source target good-fit leads (default 150 for Signal Discovery unless the
+  campaign says otherwise)
 - eligible right-content posts by lane/content type
 - reachable engagers from those posts
 - sampled ICP-fit rate as `n/N` plus an easy percentage/range
@@ -447,7 +449,7 @@ selections, headlineICPCriteria })`, then call
 targetLeadCount, targetEngagerCount, maxPostsToScrape })` for the approved
 source-capacity plan without asking for another yes/no gate. Then
 `confirm_lead_list` imports only the bounded review batch into the campaign
-table. Do not confuse the source scrape target with the 15-row review batch.
+table. Do not confuse the source-candidate target with the review-batch size.
 
 The promotion/select step is required campaign state. Use post IDs from the
 current campaign-scoped search result, not stale IDs copied from a source review
@@ -460,8 +462,8 @@ the source-list materialization.
 ```json
 import_leads({
   "campaignOfferId": "cmp_xxx",
-  "targetLeadCount": 300,
-  "targetEngagerCount": 1500,
+  "targetLeadCount": 1000,
+  "targetEngagerCount": 1000,
   "maxPostsToScrape": 5
 })
 ```
@@ -474,7 +476,7 @@ After the lead list finishes and the user confirms it looks good, call:
 confirm_lead_list({
   "campaignOfferId": "cmp_xxx",
   "jobId": "<jobId from import_leads>",
-  "targetLeadCount": 15
+  "targetLeadCount": 25
 })
 ```