@sellable/mcp 0.1.143 → 0.1.145

package/README.md

@@ -16,10 +16,11 @@ Each message gets 5+ minutes of Claude attention with deep research - no other t
 
 ### Prompt Source Of Truth
 
-There are two public Sellable entrypoints shared across hosts:
+There are three public Sellable entrypoints shared across hosts:
 
 - `sellable:create-campaign`
 - `sellable:interview`
+- `sellable:load-voice`
 
 The create-campaign public wrapper at
 `mcp/sellable/skills/create-campaign/SKILL.md` handles auth/bootstrap and loads
@@ -32,6 +33,11 @@ from:
 
 - `mcp/sellable/skills/interview/SKILL.md`
 
+The load-voice public wrapper loads the current voice/company memory for use in
+writing, application answers, posts, replies, and reviews from:
+
+- `mcp/sellable/skills/load-voice/SKILL.md`
+
 Keep `create-campaign-v2` internal. Do not advertise it as a public command.
 
 ### 1. One-Command Agent Install
@@ -127,8 +133,9 @@ The installer does the full local setup:
   `mcp__sellable__*` tools into skill sessions
 
 After the installer passes, fully quit and reopen Codex Desktop. Start a new
-thread and select `Sellable Create Campaign` / `Sellable Identity Interview`, or
-invoke `$sellable:create-campaign` / `$sellable:interview`. If the app still says
+thread and select `Sellable Create Campaign`, `Sellable Identity Interview`, or
+`Sellable Load Voice`; or invoke `$sellable:create-campaign`,
+`$sellable:interview`, or `$sellable:load-voice`. If the app still says
 `mcp__sellable__*` tools are missing after the installer passes, check that
 `~/.codex/config.toml` contains both `[marketplaces.sellable]` and
 `[plugins."sellable@sellable"]`.
@@ -141,13 +148,17 @@ Use these names consistently:
 
 - Claude Code command: `/sellable:create-campaign`
 - Claude Code command: `/sellable:interview`
+- Claude Code command: `/sellable:load-voice`
 - Codex command: `$sellable:create-campaign`
 - Codex command: `$sellable:interview`
+- Codex command: `$sellable:load-voice`
 - Codex Desktop plugin: `sellable@sellable`
 - Codex visible skill: `Sellable Create Campaign`
 - Codex visible skill: `Sellable Identity Interview`
+- Codex visible skill: `Sellable Load Voice`
 - Codex skill frontmatter name: `create-campaign`
 - Codex skill frontmatter name: `interview`
+- Codex skill frontmatter name: `load-voice`
 - MCP server name: `sellable`
 - Internal workflow prompt: `create-campaign-v2`
 
@@ -255,9 +266,11 @@ Parallel execution contract:
   through `get_source_scout_registry`. Add new scouts there first; installer,
   Codex config, Claude agent files, and prompts should consume the registry.
 - Post-find-leads scout names also come from `agents/registry.json` and are
-  exposed through `get_post_find_leads_scout_registry`. After source approval,
-  use that registry to launch the filter-leads scout and message-generation
-  scout together when the host supports real subagents.
+  exposed through `get_post_find_leads_scout_registry`. After
+  `confirm_lead_list` imports a non-empty review batch and rows are proven, use
+  that registry to launch the message-generation scout immediately when the host
+  supports real subagents. Launch the filter-leads scout later, only after the
+  user chooses filters.
 - Claude host: use the installed `source-scout-linkedin-engagement`,
   `source-scout-sales-nav`, and `source-scout-prospeo-contact` Task/Agent
   subagents for parallel lead-source scouting only when the current session

package/agents/post-find-leads-message-scout.md

@@ -136,3 +136,47 @@ Return a concise status with:
 - one rendered omit/fallback sample
 - quality-gate pass/fail summary
 - whether final template review is ready or needs revision
+
+When the parent will show the recommendation in chat, format the customer-facing
+message review as Markdown with distinct copy blocks:
+
+````markdown
+## Message Template
+
+**Subject**
+
+```text
+{{tokenized_subject}}
+```
+
+**Body**
+
+```text
+{{tokenized_message_body}}
+```
+
+## Rendered Examples
+
+### Good token fill
+
+```text
+Subject: ...
+
+Hey First,
+
+...
+```
+
+### Good omit / fallback
+
+```text
+Subject: ...
+
+Hey First,
+
+...
+```
+````
+
+Use a table for token rules. Keep explanations outside the fenced `text` blocks
+so the user can quickly see exactly what copy is being approved.

package/agents/registry.json

@@ -168,7 +168,7 @@
       "optionalProducesArtifacts": [],
       "ownership": "lead quality, false-positive patterns, keep/exclude rules, ability-to-pay checks, and production rubric translation only",
       "codex": {
-        "description": "Lead Fit Builder for campaign-backed lead filtering and rubric persistence after source approval.",
+        "description": "Lead Fit Builder for campaign-backed lead filtering and rubric persistence after review-batch import and filter approval.",
         "model": "gpt-5.5",
         "modelReasoningEffort": "high",
         "sandboxMode": "read-only",
@@ -179,7 +179,7 @@
         ]
       },
       "claude": {
-        "description": "Use proactively as Lead Fit Builder after lead source approval to persist campaign rubrics from campaign state.",
+        "description": "Use proactively as Lead Fit Builder after review-batch import and filter approval to persist campaign rubrics from campaign state.",
         "model": "inherit",
         "background": true,
         "maxTurns": 8,

package/agents/source-scout-linkedin-engagement.md

@@ -51,7 +51,8 @@ currentStep: "signal-discovery" })` before sampling so the watched Signal
    visible headline/display-name cues only. Do not enrich people during
    viability estimation.
 7. Compute capacity before recommending the source: source target good-fit
-   leads (default 300 unless the parent supplies a target), reachable engagers,
+   leads (default 150 for Signal Discovery unless the parent supplies a target),
+   reachable engagers,
    sampled ICP-fit rate as `n/N` plus an easy percentage/range, expected usable
    leads per 100 engagers before and after a conservative dedupe/cleanup
    factor, required engagers to scrape (`source target / fit rate`), average
@@ -93,8 +94,8 @@ Evidence standards:
   exist, how many sampled engagers looked in-ICP, how many good-fit prospects
   that implies per 100 engagers, how many usable prospects one right-content
   post should yield after cleanup, how many engagers must be scraped for the
-  300-good-fit source target, how many posts are needed for that source target,
+  150-good-fit source target, how many posts are needed for that source target,
   and which posts you would use. Also say the first campaign import remains the
-  bounded 15-row review batch.
+  bounded review batch.
 - If `fetch_post_engagers` is unavailable or fails, report that explicitly and mark the estimate lower-confidence.
 - Keep LinkedIn Engagement viable when selected posts can produce roughly 150+ ICP-fit warm prospects before final filtering, even if Sales Nav is more scalable.

package/agents/source-scout-prospeo-contact.md

@@ -27,7 +27,7 @@ Process:
 1. Read the campaign brief, source intake, kickoff doc, or lane prompt supplied by the parent.
 2. Identify whether this is domain/account targeting or broad persona expansion.
 3. For domain targeting, use or create the standalone `domainFilterId` before searching; never pass raw domains directly into `search_prospeo`.
-4. Run the narrowest useful Prospeo people preview and 1-2 refinements if quality or scale is unclear.
+4. Run the narrowest useful Prospeo people preview and 1-2 refinements if quality or scale is unclear. Check scale against a default 300+ good-fit target, capped at 2,500 source candidates unless the parent supplies a different target.
 5. Call out that Prospeo gives contact/account coverage but usually weaker LinkedIn intent than LinkedIn Engagement or Sales Nav activity slices.
 
 Return a concise structured result with:

package/agents/source-scout-sales-nav.md

@@ -28,8 +28,9 @@ Process:
 2. Preserve target role names with `CURRENT_TITLE` lookups; do not rely on seniority alone when the brief names concrete roles.
 3. When `lookup_sales_nav_filter` returns multiple title options, choose the closest semantic title match instead of the first result.
 4. Build a broad-but-reasonable baseline from role/title, geography, company size, industry/account context, and recent LinkedIn activity when relevant.
-5. Check scale against the source target good-fit lead count (default 300
-   unless the parent supplies a target). If raw preview volume or projected usable volume
+5. Check scale against the source target good-fit lead count (default 300+
+   for Sales Nav, capped at 2,500 source candidates unless the parent supplies a target).
+   If raw preview volume or projected usable volume
    is below target, do not present the tiny result as the scale fallback yet.
    Loosen nonessential filters in order: remove recent-activity first, widen
    adjacent title variants, widen geography/company-size constraints, and only

package/dist/tools/leads.js

@@ -14,10 +14,11 @@ const workspaceRoot = resolveWorkspaceRoot(entryDir);
 const skillsRoot = resolveSkillsDir();
 const signalProviderConfigPath = join(skillsRoot, "create-campaign/core/providers/signal-discovery.json");
 const leadImportLimitsPath = join(workspaceRoot, "lead-import-limits.json");
+const campaignSourceDefaultsPath = join(workspaceRoot, "campaign-source-defaults.json");
 const defaultLeadImportLimits = {
     apollo: { maxImportCount: 2500 },
     "sales-nav": { maxImportCount: 2500 },
-    prospeo: { maxImportCount: 10000 },
+    prospeo: { maxImportCount: 2500 },
     "signal-discovery": { maxImportCount: 2500 },
 };
 const defaultSignalDiscoveryConfig = {
@@ -25,6 +26,28 @@ const defaultSignalDiscoveryConfig = {
         promisingPostsTarget: 3,
     },
 };
+const defaultCampaignSourceDefaults = {
+    reviewBatch: {
+        defaultSize: 25,
+        minProjectedPass: 5,
+    },
+    providers: {
+        "signal-discovery": {
+            targetGoodFitLeads: 150,
+            defaultFitRate: 0.15,
+            maxSourceCandidates: 2500,
+            postCoverageBuffer: 1.2,
+        },
+        "sales-nav": {
+            targetGoodFitLeads: 300,
+            maxSourceCandidates: 2500,
+        },
+        prospeo: {
+            targetGoodFitLeads: 300,
+            maxSourceCandidates: 2500,
+        },
+    },
+};
 const prospeoFilterValueSchema = {
     type: "object",
     description: "Include/exclude list filter (values must match Prospeo enums)",
@@ -245,6 +268,39 @@ function loadLeadImportLimits() {
         return defaultLeadImportLimits;
     }
 }
+function loadCampaignSourceDefaults() {
+    if (!existsSync(campaignSourceDefaultsPath)) {
+        return defaultCampaignSourceDefaults;
+    }
+    try {
+        const raw = readFileSync(campaignSourceDefaultsPath, "utf-8");
+        const parsed = JSON.parse(raw);
+        return {
+            reviewBatch: {
+                ...defaultCampaignSourceDefaults.reviewBatch,
+                ...(parsed.reviewBatch ?? {}),
+            },
+            providers: {
+                ...defaultCampaignSourceDefaults.providers,
+                ...(parsed.providers ?? {}),
+            },
+        };
+    }
+    catch {
+        return defaultCampaignSourceDefaults;
+    }
+}
+function getSignalDiscoverySourcePlanDefaults() {
+    const defaults = loadCampaignSourceDefaults();
+    const signalDefaults = defaults.providers["signal-discovery"];
+    const sourceCandidateTarget = Math.min(Math.ceil(signalDefaults.targetGoodFitLeads / signalDefaults.defaultFitRate), signalDefaults.maxSourceCandidates);
+    return {
+        targetGoodFitLeads: signalDefaults.targetGoodFitLeads,
+        defaultFitRate: signalDefaults.defaultFitRate,
+        sourceCandidateTarget,
+        reviewBatchSize: defaults.reviewBatch.defaultSize,
+    };
+}
 function getMaxImportCount(provider) {
     const limits = loadLeadImportLimits();
     const max = limits[provider]?.maxImportCount;
@@ -443,17 +499,17 @@ function buildSourceImportWatchNarration({ provider, selectedPostCount, estimate
             : ""}`
         : `the approved ${providerLabel} source`;
     const targetDetail = typeof targetLeadCount === "number"
-        ? ` Targeting ${targetLeadCount.toLocaleString("en-US")} source leads before the bounded review batch is cloned.`
+        ? ` Targeting ${targetLeadCount.toLocaleString("en-US")} source candidates before the bounded review batch is cloned.`
         : "";
     return {
         stage: "review-batch",
         headline: provider === "signal-discovery"
-            ? "Scraping source leads from posts"
-            : "Importing source leads",
+            ? "Scraping source candidates from posts"
+            : "Importing source candidates",
         visibleState: `The browser is showing ${providerLabel} import progress for ${sourceDetail}.${targetDetail}`,
         agentIntent: "Codex is materializing the approved source into a lead list before cloning only the bounded review batch into the campaign.",
-        nextAction: "Wait for source leads, then import the 15-row review batch",
-        safety: "No enrichment, sequence, or sending starts during source import.",
+        nextAction: "Wait for source candidates, then import the review batch",
+        safety: "Import is limited to the review batch.",
     };
 }
 function buildSourceImportRecoveryWatchNarration(args) {
@@ -481,7 +537,7 @@ function buildSourceImportRecoveryWatchNarration(args) {
         nextAction: args.reason === "failed" || args.reason === "zero"
             ? "Retry the import or change the approved source"
             : "Wait again, retry readiness, or change the source",
-        safety: "No enrichment, filtering, message cells, sequence, or sending has started.",
+        safety: "The campaign is still at source recovery.",
     };
 }
 function buildFilterChoiceWatchNarration({ sourceLeadCount, reviewRowCount, }) {
@@ -497,7 +553,7 @@ function buildFilterChoiceWatchNarration({ sourceLeadCount, reviewRowCount, }) {
         visibleState: `${reviewCopy} The browser is showing the filter-choice screen and sample rows.`,
         agentIntent: "Codex is pausing on this sample because a mixed review list should be filtered before message review. Skip filters only if the visible rows already look clean.",
         nextAction: "Choose filters or skip",
-        safety: "No enrichment, filtering, Generate Message cells, sequence, or sending starts from this choice.",
+        safety: "This choice sets the filter path.",
         workerStatuses: {
             leadFitBuilder: "idle",
             messageDraftBuilder: "idle",
@@ -511,7 +567,7 @@ function buildSignalDiscoverySearchWatchNarration() {
         visibleState: "The browser is showing Signal Discovery while LinkedIn post searches run.",
         agentIntent: "Codex is searching the approved post themes and will ask before scraping selected engagers into a source list.",
         nextAction: "Review selected posts before scraping",
-        safety: "Search only stores post candidates; it does not import leads, enrich, generate messages, create a sequence, or send.",
+        safety: "Scrape approval is the next gate.",
     };
 }
 function buildSignalDiscoveryResultsWatchNarration(postsReturned) {
@@ -524,7 +580,7 @@ function buildSignalDiscoveryResultsWatchNarration(postsReturned) {
         visibleState: postCopy,
         agentIntent: "Codex is promoting the strongest posts, then it will ask before scraping the selected engager pool.",
         nextAction: "Approve selected posts before scraping",
-        safety: "Post results do not import leads, enrich, generate messages, create a sequence, or send.",
+        safety: "Scrape approval is the next gate.",
     };
 }
 function buildSelectedPostApprovalWatchNarration(selectedPostCount) {
@@ -533,10 +589,59 @@ function buildSelectedPostApprovalWatchNarration(selectedPostCount) {
         headline: "Approve selected-post scrape",
         visibleState: `${selectedPostCount.toLocaleString("en-US")} LinkedIn post${selectedPostCount === 1 ? "" : "s"} selected in Signal Discovery.`,
         agentIntent: "Codex is asking before scraping this selected engager pool into a source list and cloning only the bounded review batch into the campaign.",
-        nextAction: `Scrape ${selectedPostCount.toLocaleString("en-US")} selected post${selectedPostCount === 1 ? "" : "s"} for the bounded review batch`,
-        safety: "Post selection does not import leads, enrich, generate messages, create a sequence, or send.",
+        nextAction: `Approve scraping ${selectedPostCount.toLocaleString("en-US")} Signal Discovery post${selectedPostCount === 1 ? "" : "s"}`,
+        safety: "Scrape approval is the next gate.",
     };
 }
+function escapeMarkdownTableCell(value) {
+    return (value || "-").replace(/\|/g, "\\|").replace(/\s+/g, " ").trim();
+}
+function formatApproxInteger(value) {
+    const rounded = value >= 100 ? Math.round(value / 10) * 10 : Math.round(value);
+    return `~${rounded.toLocaleString("en-US")}`;
+}
+function buildSignalDiscoverySourceRecommendation({ selectedPosts, }) {
+    const { targetGoodFitLeads, defaultFitRate, sourceCandidateTarget, reviewBatchSize, } = getSignalDiscoverySourcePlanDefaults();
+    const selectedCount = selectedPosts.length;
+    const totalEngagement = selectedPosts.reduce((sum, post) => sum + (post.likes ?? 0) + (post.comments ?? 0), 0);
+    const tableRows = selectedPosts
+        .map((post) => {
+        const engagement = (post.likes ?? 0) + (post.comments ?? 0);
+        return `| ${escapeMarkdownTableCell(post.authorName)} | ${escapeMarkdownTableCell(post.reason)} | ${formatApproxInteger(engagement)} |`;
+    })
+        .join("\n");
+    const estimatedGoodFit = totalEngagement * defaultFitRate;
+    return `## Source Recommendation
+
+Use Signal Discovery first.
+
+**Good-fit target:** ~${targetGoodFitLeads.toLocaleString("en-US")} prospects after cleanup, enrichment, and filters<br>
+**Source-candidate plan:** scrape ~${sourceCandidateTarget.toLocaleString("en-US")} raw engagers using a conservative ${Math.round(defaultFitRate * 100)}% fit-rate assumption<br>
+**Review checkpoint:** import the first ${reviewBatchSize.toLocaleString("en-US")} leads into the campaign for fit and message review before scaling
+
+### Selected posts
+
+| Post | Why it fits | Visible engagement |
+|---|---|---:|
+${tableRows || "| Selected posts | Campaign-matched public engagement | - |"}
+
+**Total visible pool:** ${formatApproxInteger(totalEngagement)} engagers<br>
+**Estimated good-fit pool at ${Math.round(defaultFitRate * 100)}%:** ${formatApproxInteger(estimatedGoodFit)} prospects before dedupe/risk cleanup
+
+### Recommendation
+
+Approve scraping these ${selectedCount} posts.
+
+This gives enough volume to work toward ~${targetGoodFitLeads.toLocaleString("en-US")} good-fit prospects while keeping the source tied to people already engaging with the campaign's strongest public buying signals.
+
+**First pass:** build the source list, then import only the ${reviewBatchSize.toLocaleString("en-US")}-lead review batch so we can inspect fit and messages before scaling.
+
+**Fallback:** if the review batch is too vendor-heavy, agency-heavy, or off-ICP, switch to Sales Nav recent activity.
+
+Approval card should say:
+
+**Approve scraping ${selectedCount} Signal Discovery post${selectedCount === 1 ? "" : "s"}?**`;
+}
 function normalizeImportProvider(provider) {
     if (provider === "apollo-ai" || provider === "apollo")
         return "apollo";
@@ -1117,7 +1222,7 @@ export const leadToolDefinitions = [
                 },
                 targetEngagerCount: {
                     type: "number",
-                    description: "Signal Discovery: target number of post engagers to scrape. Compute from source target good-fit leads / sampled fit rate; e.g. 300 good fits at 20% fit requires about 1500 engagers. Limits selected posts before starting scrape.",
+                    description: "Signal Discovery: target number of post engagers/source candidates to scrape. Compute from source target good-fit leads / sampled fit rate; e.g. 150 good fits at 15% fit requires about 1000 engagers. Limits selected posts before starting scrape.",
                 },
                 maxPostsToScrape: {
                     type: "number",
@@ -2726,12 +2831,61 @@ export async function selectPromisingPosts(input) {
         currentStep: "signal-discovery",
         watchNarration: buildSelectedPostApprovalWatchNarration(selectionResult.selectedCount),
     });
+    let sourceRecommendation = "";
+    try {
+        const tabsResponse = await api.get(`/api/v3/campaigns/${campaignOfferId}/signal-discovery/tabs`);
+        const reasonsByPostId = new Map(selections.map((selection) => [selection.postId, selection.reason]));
+        const selectedByUrl = new Map();
+        for (const tab of tabsResponse.tabs ?? []) {
+            for (const post of tab.posts ?? []) {
+                if (!post.isSelected)
+                    continue;
+                const urlKey = normalizePostUrl(post.postUrl ?? undefined) ||
+                    post.postUrl ||
+                    post.id ||
+                    "";
+                if (!urlKey)
+                    continue;
+                const directReason = post.id ? reasonsByPostId.get(post.id) : undefined;
+                const urlReason = post.postUrl && reasonsByPostId.get(post.postUrl)
+                    ? reasonsByPostId.get(post.postUrl)
+                    : undefined;
+                selectedByUrl.set(urlKey, {
+                    authorName: post.authorName,
+                    reason: directReason ||
+                        urlReason ||
+                        (post.postContent ? truncate(post.postContent, 70) : null),
+                    likes: post.likes,
+                    comments: post.comments,
+                });
+            }
+        }
+        sourceRecommendation = buildSignalDiscoverySourceRecommendation({
+            selectedPosts: Array.from(selectedByUrl.values()),
+        });
+    }
+    catch {
+        const { reviewBatchSize } = getSignalDiscoverySourcePlanDefaults();
+        sourceRecommendation = `## Source Recommendation
+
+Use Signal Discovery first.
+
+**Recommendation:** approve scraping the ${selectionResult.selectedCount} selected Signal Discovery post${selectionResult.selectedCount === 1 ? "" : "s"}.
+
+**First pass:** build the source list, then import only the ${reviewBatchSize.toLocaleString("en-US")}-lead review batch so we can inspect fit and messages before scaling.
+
+Approval card should say:
+
+**Approve scraping ${selectionResult.selectedCount} Signal Discovery post${selectionResult.selectedCount === 1 ? "" : "s"}?**`;
+    }
     return {
         success: true,
         selectedCount: selectionResult.selectedCount,
         unselectedCount: selectionResult.unselectedCount,
         criteriaCount: selectionResult.criteriaCount,
-        message: `Selected ${selectionResult.selectedCount} posts with ${selectionResult.criteriaCount} ICP criteria (persisted). Criteria are already saved to the campaign. Ask the user to approve scraping these selected posts before calling import_leads; the earlier source-plan approval only authorized search/scouting.`,
+        message: `${sourceRecommendation}
+
+Selected ${selectionResult.selectedCount} posts with ${selectionResult.criteriaCount} ICP criteria persisted. Ask the user to approve the specific scraping action before calling import_leads; the earlier source-plan approval only authorized search/scouting.`,
     };
 }
 export async function setHeadlineICPCriteria(input) {

package/dist/tools/processing.d.ts

@@ -60,6 +60,7 @@ export type CampaignRubricsResponse = {
     enableICPFilters: boolean | null;
     currentStep?: string | null;
     workflowTableId: string | null;
+    hasApprovedMessageTemplate?: boolean | null;
     rubrics: RubricItem[];
 };
 export declare function fetchCampaignRubrics(campaignOfferId: string): Promise<CampaignRubricsResponse>;

package/dist/tools/prompts.js

@@ -339,9 +339,9 @@ export function getPostFindLeadsScoutRegistry() {
             reusePolicy: "The first completed Message Draft Builder recommendation remains the default review candidate. Later Lead Fit Builder, Filter Leads, enrichment, or rubric completion may make an enriched rewrite available, but does not automatically retry or replace the initial draft unless campaign/brief/source/list/table/review-batch identity mismatches or the initial output failed.",
         },
         usage: {
-            codex: "After the user approves or auto-confirms the lead source, spawn both returned scout `name` values in one assistant turn only when the current Codex host exposes those custom agents.",
-            claude: "After lead source approval, invoke both returned Task/Agent subagents in one assistant message only when the current Claude session lists those agents, so filter-leads and message generation run concurrently.",
-            parentThreadRule: "Named agents are optional acceleration. If they are absent, do not customer-surface install status; the main thread still orchestrates filter and message branches from CampaignOffer state, selected source state, workflowTableId, and imported review-batch rows. Local markdown/json files are not normal-path inputs. The message drafting branch must load the full generate-messages prompt, must read live campaign/review-batch state through scoped MCP/product tools, and must reject mismatched selectedLeadListId/workflowTableId/campaign/workspace input. Template approval in the parent thread uses the create-campaign-v2 message-review safety gate; do not load the full generate-messages prompt for approval-only review. Join before message review. Do not automatically rerun Message Draft Builder after filters/enrichment finish; show the initial draft by default and offer an enriched rewrite only with explicit user opt-in.",
+            codex: "After confirm_lead_list imports a non-empty bounded review batch and get_rows_minimal proves rows for workflowTableId, spawn the returned Message Draft Builder scout name immediately when the current Codex host exposes that custom agent. Spawn the Lead Fit Builder scout only after the user chooses filters.",
+            claude: "After confirm_lead_list imports a non-empty bounded review batch and get_rows_minimal proves rows for workflowTableId, invoke the returned Message Draft Builder Task/Agent immediately when the current Claude session lists that agent. Invoke the Lead Fit Builder only after the user chooses filters.",
+            parentThreadRule: "Named agents are optional acceleration. If they are absent, do not customer-surface install status; the main thread still orchestrates filter and message branches from CampaignOffer state, selected source state, workflowTableId, and imported review-batch rows. Local markdown/json files are not normal-path inputs. The message drafting branch starts at the review-batch import gate, must load the full generate-messages prompt, must read live campaign/review-batch state through scoped MCP/product tools, and must reject mismatched selectedLeadListId/workflowTableId/campaign/workspace input. Template approval in the parent thread uses the create-campaign-v2 message-review safety gate; do not load the full generate-messages prompt for approval-only review. Join before message review. Do not automatically rerun Message Draft Builder after filters/enrichment finish; show the initial draft by default and offer an enriched rewrite only with explicit user opt-in.",
         },
     };
 }

package/dist/tools/rubrics.js

@@ -86,21 +86,26 @@ async function fetchCampaignOffer(campaignOfferId) {
         workflowTableId: v3.workflowTableId,
         enableICPFilters: v3.enableICPFilters,
         currentStep: v3.currentStep,
+        hasApprovedMessageTemplate: v3.hasApprovedMessageTemplate,
         leadScoringRubrics: v3.rubrics,
     };
 }
 const filterLeadsReadyWatchNarration = {
     stage: "fit-message",
-    headline: "Filter Leads is ready",
-    visibleState: "Filters are enabled, rubrics are saved, and the browser is showing Filter Leads.",
-    agentIntent: "Codex is waiting for the approved message template before running enrichment, filtering, or message cells.",
-    nextAction: "Review or approve the message template before filtering",
+    headline: "Filter rules saved",
+    visibleState: "The browser is staying on Filter Leads with the fit rules saved.",
+    agentIntent: "Codex is preparing the message template for approval before enrichment, filtering, or message cells run.",
+    nextAction: "Review the message template in chat",
     progressLabel: "Fit + message",
-    safety: "Saved rules do not start enrichment, filtering, Generate Message cells, sequence setup, or sending.",
+    safety: "Template approval comes before row execution.",
 };
-function shouldMoveToFilterLeads(currentStep) {
+function shouldMoveToFilterLeads(campaign) {
+    const currentStep = campaign.currentStep;
     if (!currentStep)
         return true;
+    if (currentStep === "messages") {
+        return campaign.hasApprovedMessageTemplate === false;
+    }
     return [
         "confirm-lead-list",
         "filter-choice",
@@ -109,11 +114,11 @@ function shouldMoveToFilterLeads(currentStep) {
         "apply-icp-rubric",
     ].includes(currentStep);
 }
-function buildEnableIcpFiltersPayload(currentStep) {
+function buildEnableIcpFiltersPayload(campaign) {
     const payload = {
         enableICPFilters: true,
     };
-    if (shouldMoveToFilterLeads(currentStep)) {
+    if (shouldMoveToFilterLeads(campaign)) {
         payload.currentStep = "apply-icp-rubric";
         payload.watchNarration = filterLeadsReadyWatchNarration;
     }
@@ -568,7 +573,7 @@ export async function saveRubrics(input) {
     let currentStepSet = false;
     try {
         const api = getApi();
-        const payload = buildEnableIcpFiltersPayload(campaign.currentStep);
+        const payload = buildEnableIcpFiltersPayload(campaign);
         await api.put(`/api/v2/campaign-offers/${input.campaignOfferId}`, payload);
         enableICPFiltersSet = true;
         currentStepSet = payload.currentStep === "apply-icp-rubric";

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@sellable/mcp",
-  "version": "0.1.143",
+  "version": "0.1.145",
   "type": "module",
   "description": "Sellable MCP server for Claude Code and Codex campaign workflows",
   "main": "dist/index.js",