@productbrain/mcp 0.0.1-beta.4881 → 0.0.1-beta.4887

package/dist/{chunk-EW72I5PP.js → chunk-NGTA422J.js}

@@ -15571,7 +15571,7 @@ import { ResourceTemplate } from "@modelcontextprotocol/sdk/server/mcp.js";
 // src/resources/agentCheatsheet.ts
 var AGENT_CHEATSHEET = `# Product Brain \u2014 Agent Cheatsheet
 
-## Core Tools (Serves 13 default compound tools; 3 more (chain, chain-review, map) behind PB_MODULES=gitchain)
+## Core Tools (Serves 13 default compound tools; 2 more (chain, chain-review) behind PB_MODULES=gitchain \u2014 map retired, L29)
 | Tool | Purpose | Key params |
 |---|---|---|
 | \`orient\` | Workspace context, governance, active work packages, session start | \`action\`: start, task, record-activation |
@@ -15592,7 +15592,7 @@ var AGENT_CHEATSHEET = `# Product Brain \u2014 Agent Cheatsheet
 GLO (glossary), BR (business-rules), PRI (principles), STD (standards),
 DEC (decisions), STR (strategy), TEN (tensions), FEAT (features),
 BET (bets), INS (insights), ARCH (architecture), TEAM (teams),
-ROL (roles), MAP (maps), MTRC (tracking-events), ST (semantic-types)
+ROL (roles), MTRC (tracking-events), ST (semantic-types)
 
 ## Valid Relation Types (24)
 informs, governs, surfaces_tension_in, defines_term_for, belongs_to,
@@ -15668,7 +15668,7 @@ These are principles, standards, and rules your team has accepted onto the Chain
     );
   }
   sections.push(
-    "## Core Product Architecture\nThe Chain is a versioned, connected, compounding knowledge base \u2014 the SSOT for a product team.\nEverything on the Chain is one of **three primitives**:\n\n- **Entry** (atom) \u2014 a discrete knowledge unit: target audience, business rule, glossary term, metric, tension. Lives in a collection.\n- **Process** (authored) \u2014 a narrative chain with named links (Strategy Coherence, IDM Proposal). Content is inline text. Collection: `chains`.\n- **Map** (composed) \u2014 a framework assembled by reference (Lean Canvas, Empathy Map). Slots point to ingredient entries. Collection: `maps`.\n\nEntries are atoms. Processes author narrative from atoms. Maps compose frameworks from atoms.\nAll three share the same versioning, branching, gating, and relation system.\n\nThe Chain compounds: each new relation makes entries discoverable from more starting points \u2192 context gathering returns richer results \u2192 AI processes have more context \u2192 quality scores improve \u2192 next accept is smarter than the last."
+    "## Core Product Architecture\nThe Chain is a versioned, connected, compounding knowledge base \u2014 the SSOT for a product team.\nEverything on the Chain is one of **two primitives**:\n\n- **Entry** (atom) \u2014 a discrete knowledge unit: target audience, business rule, glossary term, metric, tension. Lives in a collection.\n- **Process** (authored) \u2014 a narrative chain with named links (Strategy Coherence, IDM Proposal). Content is inline text. Collection: `chains`.\n\nEntries are atoms. Processes author narrative from atoms.\nBoth share the same versioning, branching, gating, and relation system.\n\nThe Chain compounds: each new relation makes entries discoverable from more starting points \u2192 context gathering returns richer results \u2192 AI processes have more context \u2192 quality scores improve \u2192 next accept is smarter than the last."
   );
   sections.push(
     "## Architecture\n```\nCursor (stdio) \u2192 MCP Server (mcp-server/src/index.ts)\n  \u2192 POST /api/aki with Bearer token\n  \u2192 Convex HTTP Action (convex/http.ts)\n  \u2192 internalQuery / internalMutation (convex/agentKnowledge/*)\n  \u2192 Convex DB (workspace-scoped)\n```\nSecurity: API key auth on every request, workspace-scoped data, internal functions blocked from external clients.\nKey files: `packages/mcp-server/src/client.ts` (HTTP client + audit), `convex/schema.ts` (schema)."
@@ -15681,7 +15681,7 @@ These are principles, standards, and rules your team has accepted onto the Chain
     sections.push(
       `## Data Model (${collections.length} collections)
 Unified entries model: collections define field schemas, entries hold data in a flexible \`data\` field.
-The \`data\` field is polymorphic: plain fields for generic entries, \`ChainData\` (chainTypeId + links) for processes, \`MapData\` (templateId + slots) for maps.
+The \`data\` field is polymorphic: plain fields for generic entries, \`ChainData\` (chainTypeId + links) for processes.
 Tags for filtering (e.g. \`severity:high\`), relations via \`entryRelations\`, versions via \`entryVersions\`, labels via \`labels\` + \`entryLabels\`.
 
 ` + collList + "\n\nUse `collections action=list` for field schemas, `entries action=get` for full records."
@@ -15709,10 +15709,10 @@ Implementation: \`src/lib/analytics.ts\`. Workspace-scoped events MUST use \`wit
 Browse: \`entries action=list collection=tracking-events\`.`
   );
   sections.push(
-    "## Knowledge Graph\nEntries are connected via typed relations (`entryRelations` table). Relations are bidirectional and collection-agnostic \u2014 any entry can link to any other entry.\n\n**Recommended relation types** (extensible \u2014 any string accepted):\n- `governs` \u2014 a rule constrains behavior of a feature\n- `defines_term_for` \u2014 a glossary term is canonical vocabulary for a feature/area\n- `belongs_to` \u2014 a feature belongs to a product area or parent concept (WP-641 typed successors: `anchored_to`, `authority_domain`)\n- `informs` \u2014 a decision or insight informs a feature\n- `fills_slot` \u2014 an ingredient entry fills a slot in a map\n- `surfaces_tension_in` \u2014 a tension exists within a feature area\n- `related_to`, `depends_on`, `replaces`, `conflicts_with`, `references`, `confused_with`\n\nEach relation type is defined as a glossary entry (prefix `GT-REL-*`) to prevent terminology drift. **Workflow-owned, NOT generic:** `domain_routed_to` (a question routed to a domain) \u2014 use the `question` tool (`action=create`/`adopt`, `domain=<slug>`), never `relations action=create`; the generic door has no replace/re-judge logic, so a raw write can leave duplicate routing edges and a stale suggested owner.\n\n**Tools:**\n- `context action=gather` \u2014 get the full context around any entry (multi-hop graph traversal)\n- `relations action=suggest` \u2014 discover potential connections for an entry\n- `relations action=create` \u2014 create a typed link between two entries\n- `relations action=find` \u2014 list direct relations for an entry\n\n**Convention:** When creating or updating entries in governed collections, always use `relations action=suggest` to discover and create relevant relations."
+    "## Knowledge Graph\nEntries are connected via typed relations (`entryRelations` table). Relations are bidirectional and collection-agnostic \u2014 any entry can link to any other entry.\n\n**Recommended relation types** (extensible \u2014 any string accepted):\n- `governs` \u2014 a rule constrains behavior of a feature\n- `defines_term_for` \u2014 a glossary term is canonical vocabulary for a feature/area\n- `belongs_to` \u2014 a feature belongs to a product area or parent concept (WP-641 typed successors: `anchored_to`, `authority_domain`)\n- `informs` \u2014 a decision or insight informs a feature\n- `surfaces_tension_in` \u2014 a tension exists within a feature area\n- `related_to`, `depends_on`, `replaces`, `conflicts_with`, `references`, `confused_with`\n\nEach relation type is defined as a glossary entry (prefix `GT-REL-*`) to prevent terminology drift. **Workflow-owned, NOT generic:** `domain_routed_to` (a question routed to a domain) \u2014 use the `question` tool (`action=create`/`adopt`, `domain=<slug>`), never `relations action=create`; the generic door has no replace/re-judge logic, so a raw write can leave duplicate routing edges and a stale suggested owner.\n\n**Tools:**\n- `context action=gather` \u2014 get the full context around any entry (multi-hop graph traversal)\n- `relations action=suggest` \u2014 discover potential connections for an entry\n- `relations action=create` \u2014 create a typed link between two entries\n- `relations action=find` \u2014 list direct relations for an entry\n\n**Convention:** When creating or updating entries in governed collections, always use `relations action=suggest` to discover and create relevant relations."
   );
   sections.push(
-    "## Creating Knowledge\n**Entries:** Use `capture` as the primary tool for creating new entries. It handles the full workflow in one call:\n1. Creates the entry with collection-aware defaults (auto-fills dates, infers domains, sets priority)\n2. Auto-links related entries from across the chain (up to 5 confident matches)\n3. Returns a quality scorecard (X/10) with actionable improvement suggestions\n\n**Smart profiles** exist for: `tensions`, `business-rules`, `glossary`, `decisions`, `features`, `audiences`, `strategy`, `standards`, `maps`, `chains`, `tracking-events`.\nAll other collections use the `ENT-{random}` fallback profile.\n\n**Processes:** Use `chain action=create` with a template (e.g., `strategy-coherence`, `idm-proposal`). Fill links with `chain action=edit`.\n\n**Prompts** (invoke via MCP prompt protocol):\n- `name-check` \u2014 verify a name against glossary conventions\n- `draft-decision-record` \u2014 scaffold a decision entry\n- `review-against-rules` \u2014 check work against business rules for a domain\n- `draft-rule-from-context` \u2014 create a new business rule from context\n\nUse `quality action=check` to score existing entries retroactively.\nUse `entries action=update` for post-creation adjustments (status changes, field updates, deprecation)."
+    "## Creating Knowledge\n**Entries:** Use `capture` as the primary tool for creating new entries. It handles the full workflow in one call:\n1. Creates the entry with collection-aware defaults (auto-fills dates, infers domains, sets priority)\n2. Auto-links related entries from across the chain (up to 5 confident matches)\n3. Returns a quality scorecard (X/10) with actionable improvement suggestions\n\n**Smart profiles** exist for: `tensions`, `business-rules`, `glossary`, `decisions`, `features`, `audiences`, `strategy`, `standards`, `chains`, `tracking-events`.\nAll other collections use the `ENT-{random}` fallback profile.\n\n**Processes:** Use `chain action=create` with a template (e.g., `strategy-coherence`, `idm-proposal`). Fill links with `chain action=edit`.\n\n**Prompts** (invoke via MCP prompt protocol):\n- `name-check` \u2014 verify a name against glossary conventions\n- `draft-decision-record` \u2014 scaffold a decision entry\n- `review-against-rules` \u2014 check work against business rules for a domain\n- `draft-rule-from-context` \u2014 create a new business rule from context\n\nUse `quality action=check` to score existing entries retroactively.\nUse `entries action=update` for post-creation adjustments (status changes, field updates, deprecation)."
   );
   sections.push(
     "## Where to Go Next\n- **Create entry** \u2192 `capture` tool (auto-links + quality score in one call)\n- **Full context** \u2192 `context action=gather` (by entry ID or task description)\n- **Discover links** \u2192 `relations action=suggest`\n- **Quality audit** \u2192 `quality action=check`\n- **Terminology** \u2192 `name-check` prompt or `productbrain://terminology` resource\n- **Schema details** \u2192 `productbrain://collections` resource or `collections action=list`\n- **Labels** \u2192 `productbrain://labels` resource or `collections action=label-list`\n- **Any collection** \u2192 `productbrain://{slug}/entries` resource\n- **Log a decision** \u2192 `draft-decision-record` prompt\n- **Growth funnel** \u2192 `productbrain://growth-funnel` resource or `entries action=list collection=strategy status=draft`\n- **Audiences** \u2192 `productbrain://audiences/entries` resource or `entries action=list collection=audiences`\n- **Session identity** \u2192 `workspace action=whoami`\n- **Health check** \u2192 `workspace action=check`\n- **Debug MCP calls** \u2192 `workspace action=audit`"
@@ -16495,4 +16495,4 @@ export {
   createProductBrainServer,
   initFeatureFlags
 };
-//# sourceMappingURL=chunk-EW72I5PP.js.map
+//# sourceMappingURL=chunk-NGTA422J.js.map