@rubytech/create-maxy-code 0.1.158 → 0.1.161

package/payload/platform/neo4j/schema.cypher

@@ -219,9 +219,9 @@ OPTIONS {
 // shared label — a WhatsApp export is a transcript, not a knowledge
 // document. Task 397 split the conversation path off into its own
 // `:ConversationArchive` parent; `:KnowledgeDocument` continues to mean
-// narrative documents only (PDF/web/Obsidian/Notion/contract attachment +
-// Gmail thread under the legacy `source:'email'` shape until Task 397's
-// email follow-up retargets it).
+// narrative documents only (PDF/web/Obsidian/Notion/contract attachment).
+// Email threads moved onto `:ConversationArchive {source:'email'}` in
+// Task 426.
 //
 // KnowledgeDocument   -[HAS_SECTION]-> Section
 // KnowledgeDocument   -[PARTY]-> (Person|Organization)        (contracts only)
@@ -233,9 +233,9 @@ OPTIONS {
 //
 // Two writer paths (see platform/plugins/memory/mcp/src/tools/memory-ingest.ts):
 //   - Document path — MERGE :KnowledgeDocument keyed on `attachmentId`
-//     (URL, file path, obsidian page, contract attachment, Gmail thread,
-//     etc.). First-ingest re-writes sections per run; anchor edges to
-//     subject nodes; orphan-candidate related entities.
+//     (URL, file path, obsidian page, contract attachment, etc.).
+//     First-ingest re-writes sections per run; anchor edges to subject
+//     nodes; orphan-candidate related entities.
 //   - Conversation path — MERGE :ConversationArchive keyed on
 //     `conversationIdentity` (= sha256(accountId + ":" +
 //     sortedParticipantElementIds)). Source-agnostic across WhatsApp,
@@ -365,17 +365,32 @@ OPTIONS {
 //
 // Label union — every operator-meaningful label:
 //   - Business identity: LocalBusiness, Service, PriceSpecification, OpeningHoursSpecification, Organization
+//   - Estate-agent ontology (Task 358): Listing, Property, Viewing, Offer, PostalAddress
 //   - People: Person, UserProfile, Preference, AdminUser, AccessGrant
 //   - Knowledge: KnowledgeDocument, ConversationArchive (Task 397), Section, Chunk (legacy),
 //     DigitalDocument, CreativeWork, Question, FAQPage, DefinedTerm, Review, ImageObject
 //   - Conversational: Conversation, AdminConversation, PublicConversation, Message,
 //     UserMessage, AssistantMessage, ToolCall
-//   - Tasks/projects/events: Task, Project, Event
+//   - Tasks/projects/events: Task, Project, Event, Meeting, Attachment
 //   - Workflows: Workflow, WorkflowStep, WorkflowRun, StepResult
 //   - Email: EmailAccount (config node only — email threads land on
-//     :KnowledgeDocument {source:'email'} since Task 321)
+//     :ConversationArchive {source:'email'} since Task 426)
 //   - CV/career sublabels: Position, Credential
 //   - Public agents: Agent (projection of file-based public agents)
+//   - Ideas + concepts: Idea, Concept (admin-twin compiledTruth subjects)
+//   - Voice mirror (Task 356): VoiceProfile, VoiceEdit
+//   - Authored content / audit: SocialPost, AEOAudit, Page
+//   - Compiled truth (Task 306/391/392): TimelineEvent, CompiledTruthRevision,
+//     Report, OrphanCandidate, CitationProposal
+//
+// Excluded by design (carry no operator-meaningful free text, asserted by
+// the doctrine test's EXCLUDED_FROM_INDEX set):
+//   - Trashed, GraphPreference                            — system flags / UI metadata
+//   - Email (legacy, retired Task 321; further retired Task 426) — replaced by :ConversationArchive {source:'email'}
+//   - Session, AnonVisitor, PageView, Click,
+//     ScrollMilestone, Recommendation                     — visitor analytics (Task 357)
+//   - Component                                           — admin-resume rehydration JSON
+//   - CloudflareTunnel, CloudflareHostname                — legacy projection (Task 288)
 //
 // Property union — every textual property the schema's writers assign:
 //   - Generic: name, title, summary, body, content, text, description, headline, abstract,
@@ -417,13 +432,16 @@ DROP INDEX entity_search_public IF EXISTS;
 
 CREATE FULLTEXT INDEX entity_search_admin IF NOT EXISTS
 FOR (n:LocalBusiness|Service|PriceSpecification|OpeningHoursSpecification|Organization
+    |Listing|Property|Viewing|Offer|PostalAddress
     |Person|UserProfile|Preference|AdminUser|AccessGrant
     |KnowledgeDocument|ConversationArchive|Section|Chunk|DigitalDocument|CreativeWork|Question|FAQPage|DefinedTerm|Review|ImageObject
     |Conversation|AdminConversation|PublicConversation|Message|UserMessage|AssistantMessage|ToolCall
-    |Task|Project|Event
+    |Task|Project|Event|Meeting|Attachment
     |Workflow|WorkflowStep|WorkflowRun|StepResult
     |EmailAccount
-    |Position|Credential|Agent|Idea)
+    |Position|Credential|Agent|Idea|Concept
+    |VoiceProfile|VoiceEdit|SocialPost|AEOAudit|Page
+    |TimelineEvent|CompiledTruthRevision|Report|OrphanCandidate|CitationProposal)
 ON EACH [n.name, n.firstName, n.lastName, n.givenName, n.familyName,
          n.title, n.currentTitle, n.summary, n.body, n.content, n.text, n.description, n.headline, n.abstract,
          n.email, n.note, n.label, n.value, n.message, n.preview, n.tagline,
@@ -434,13 +452,16 @@ ON EACH [n.name, n.firstName, n.lastName, n.givenName, n.familyName,
 
 CREATE FULLTEXT INDEX entity_search_public IF NOT EXISTS
 FOR (n:LocalBusiness|Service|PriceSpecification|OpeningHoursSpecification|Organization
+    |Listing|Property|Viewing|Offer|PostalAddress
     |Person|UserProfile|Preference|AdminUser|AccessGrant
     |KnowledgeDocument|ConversationArchive|Section|Chunk|DigitalDocument|CreativeWork|Question|FAQPage|DefinedTerm|Review|ImageObject
     |Conversation|AdminConversation|PublicConversation|Message|UserMessage|AssistantMessage|ToolCall
-    |Task|Project|Event
+    |Task|Project|Event|Meeting|Attachment
     |Workflow|WorkflowStep|WorkflowRun|StepResult
     |EmailAccount
-    |Position|Credential|Agent|Idea)
+    |Position|Credential|Agent|Idea|Concept
+    |VoiceProfile|VoiceEdit|SocialPost|AEOAudit|Page
+    |TimelineEvent|CompiledTruthRevision|Report|OrphanCandidate|CitationProposal)
 ON EACH [n.name, n.firstName, n.lastName, n.givenName, n.familyName,
          n.title, n.currentTitle, n.summary, n.body, n.content, n.text, n.description, n.headline, n.abstract,
          n.email, n.note, n.label, n.value, n.message, n.preview, n.tagline,
@@ -516,15 +537,31 @@ OPTIONS {
 // Neo4j labels are set-semantic; the sublabel is added at ON CREATE SET
 // time in `ensureConversation` so the node carries both `:Conversation`
 // and the sublabel. No new constraints — the existing
-// conversation_id_unique constraint applies to Conversations regardless
-// of sublabel. The sublabel drives /graph canvas colour resolution and
-// filter popover chips per agentType without schema fragmentation.
+// conversation_session_id_unique constraint applies to Conversations
+// regardless of sublabel. The sublabel drives /graph canvas colour
+// resolution and filter popover chips per agentType without schema
+// fragmentation.
 // Pre-Task-633 Conversations stay unlabelled and render in the base
 // :Conversation colour (sublabel backfill is a separate task).
+//
+// Task 429 — kill `conversationId`; `sessionId` is canonical. One-shot
+// migration: copy any legacy `conversationId` property onto `sessionId`
+// and remove the legacy property. Idempotent via WHERE guard: re-runs
+// against already-migrated rows are no-ops. The legacy
+// `conversation_id_unique` constraint is dropped by name; the new
+// uniqueness constraint is `conversation_session_id_unique` on
+// `:Conversation.sessionId` (disambiguated from the :Session
+// constraint that legitimately owns `session_id_unique`).
 // ----------------------------------------------------------
 
-CREATE CONSTRAINT conversation_id_unique IF NOT EXISTS
-FOR (c:Conversation) REQUIRE c.conversationId IS UNIQUE;
+DROP CONSTRAINT conversation_id_unique IF EXISTS;
+
+MATCH (c:Conversation) WHERE c.conversationId IS NOT NULL
+SET c.sessionId = coalesce(c.sessionId, c.conversationId)
+REMOVE c.conversationId;
+
+CREATE CONSTRAINT conversation_session_id_unique IF NOT EXISTS
+FOR (c:Conversation) REQUIRE c.sessionId IS UNIQUE;
 
 CREATE CONSTRAINT conversation_session_key_unique IF NOT EXISTS
 FOR (c:Conversation) REQUIRE c.sessionKey IS UNIQUE;
@@ -574,7 +611,7 @@ OPTIONS {
 //
 // Chain-fork invariant: the write path in
 // platform/ui/app/lib/neo4j-store.ts#persistMessage holds a
-// per-conversationId async mutex around the Cypher that observes
+// per-sessionId async mutex around the Cypher that observes
 // the tail and CREATEs the new [:NEXT] edge. Without this, two
 // concurrent writes under READ_COMMITTED both see the same tail
 // and each CREATE an outgoing NEXT → two edges from one node.
@@ -590,10 +627,10 @@ CREATE CONSTRAINT message_id_unique IF NOT EXISTS
 FOR (m:Message) REQUIRE m.messageId IS UNIQUE;
 
 // Message.sessionKey is the canonical conversation pointer.
-// The legacy m.conversationId index stays until the final cutover, after the
-// full code-rename slice lands and no path reads m.conversationId.
+// The legacy m.sessionId index stays until the final cutover, after the
+// full code-rename slice lands and no path reads m.sessionId.
 CREATE INDEX message_conversation IF NOT EXISTS
-FOR (m:Message) ON (m.conversationId);
+FOR (m:Message) ON (m.sessionId);
 
 CREATE INDEX message_session_key IF NOT EXISTS
 FOR (m:Message) ON (m.sessionKey);
@@ -798,24 +835,19 @@ OPTIONS {
 // ----------------------------------------------------------
 
 // ----------------------------------------------------------
-// Email threads — :KnowledgeDocument {source:'email'}
+// Email threads — :ConversationArchive {source:'email'}
 //
-// Task 321 retired the legacy :Email label. Every inbound (and outbound)
-// email lands as one :KnowledgeDocument per thread, keyed on
-// attachmentId = sha256("email:<accountId>:<rootMessageId>"). The body
-// is the verbatim chronological transcript; per-message envelopes live
-// in a JSON-serialised `messages` property; address parties hang off
-// the document as :Person edges (FROM/TO/CC). The email-fetch script
-// writes them via storeEmailMessage in plugins/email/mcp/src/lib/graph.ts.
+// Task 426 retargeted email threads off :KnowledgeDocument onto the
+// conversation-archive pipeline. The email plugin's email-ingest tool
+// dispatches `conversation-archive-ingest.sh --source email` once per
+// thread; each invocation MERGEs one :ConversationArchive
+// (conversationIdentity-keyed, source-scoped) and writes :Section
+// chunks via :HAS_SECTION + :NEXT.
 //
-// No :Email constraints or indexes are declared. The general
-// KnowledgeDocument constraints + indexes (above) already cover
-// uniqueness, accountId lookup, and embedding search. Email-specific
+// No email-specific constraints or indexes are declared. The general
+// ConversationArchive constraints + indexes (above) cover uniqueness,
+// accountId lookup, and Section embedding search. Email-specific
 // filtering (source='email') uses the generic accountId + source path.
-//
-// Relationships:
-//   (KnowledgeDocument {source:'email'})-[:RECEIVED_BY]->(EmailAccount)
-//   (KnowledgeDocument {source:'email'})-[:FROM|:TO|:CC]->(Person)
 // ----------------------------------------------------------
 
 // ----------------------------------------------------------
@@ -992,7 +1024,7 @@ FOR (tc:ToolCall) ON (tc.accountId, tc.startedAt);
 //
 // Properties:
 //   componentId  — UUID, primary key
-//   conversationId / accountId / messageId — denormalised for
+//   sessionId / accountId / messageId — denormalised for
 //     account-scoped reads and cross-account isolation audits
 //   name         — component name (document-editor, action-buttons,
 //                   single-select, etc.)
@@ -1033,7 +1065,7 @@ FOR (c:Component) ON (c.accountId);
 // Properties:
 //   attachmentId   — UUID assigned at storeAttachment time; matches
 //                     the on-disk directory + Neo4j node id.
-//   conversationId — denormalised for filtered scans + cascade delete.
+//   sessionId — denormalised for filtered scans + cascade delete.
 //   accountId      — account-isolation. Reads MUST scope by this.
 //   messageId      — back-pointer to parent :Message.
 //   filename       — original upload name (kept for chip render).
@@ -1258,9 +1290,9 @@ FOR (a:PostalAddress) ON (a.postalCode);
 // the next distillation.
 //
 // `authorshipMode` is an enum property on content-bearing labels
-// (`:KnowledgeDocument` (includes email threads under the legacy
-// `source:'email'` shape), `:ConversationArchive` (Task 397 — every chat
-// transcript), `:Message`, `:SocialPost`) with
+// (`:KnowledgeDocument`, `:ConversationArchive` (Task 397 — every chat
+// transcript, including email threads under `source:'email'` post-Task-426),
+// `:Message`, `:SocialPost`) with
 // values `human-only | human-led-agent-assisted | agent-led-human-reviewed
 // | agent-only | unknown`. Legacy nodes default to `unknown` until
 // reviewed via the voice-mirror backfill flow. The index lets the
@@ -1534,9 +1566,10 @@ CREATE INDEX local_business_compiled_truth_public_updated_at IF NOT EXISTS
 FOR (b:LocalBusiness) ON (b.compiledTruthPublicUpdatedAt);
 
 // ----------------------------------------------------------
-// :CompiledTruthRevision (Task 391) — audit history of every
-// `compiledTruth` value an entity has ever held. Captured in the
-// same transaction as the rewriter SET so revision history can
+// :CompiledTruthRevision (Task 391, twin discriminator added in
+// Task 436) — audit history of every `compiledTruth` /
+// `compiledTruthPublic` value an entity has ever held. Captured in
+// the same transaction as the writer's SET so revision history can
 // never desync from the live property. Pruned by dream-cycle
 // phase 5 (keep newest 20 per entity) — never written-and-pruned
 // in the same TX as the main rewrite.
@@ -1551,9 +1584,13 @@ FOR (b:LocalBusiness) ON (b.compiledTruthPublicUpdatedAt);
 //   MERGE (e)-[:HAS_REVISION]->(r:CompiledTruthRevision {revisionId: $id})
 // Never label-only MERGE on :CompiledTruthRevision.
 //
+// `twin` enum: 'compiledTruth' (private/admin twin) or
+// 'compiledTruthPublic' (public twin). One revision row per twin
+// per write; the specialist calls `memory-update` once per twin.
+//
 // `source` enum (exact, no catch-all):
-//   'rewriter' — compiled-truth-rewriter library wrote the truth.
-//   'operator' — operator overrode via a sanctioned write route.
+//   'rewriter' — compiled-truth-rewriter specialist wrote the truth.
+//   'operator' — operator hint overrode the rewriter for this twin.
 // ----------------------------------------------------------
 
 CREATE CONSTRAINT compiled_truth_revision_id_unique IF NOT EXISTS
@@ -1565,6 +1602,9 @@ FOR (r:CompiledTruthRevision) ON (r.accountId);
 CREATE INDEX compiled_truth_revision_written_at IF NOT EXISTS
 FOR (r:CompiledTruthRevision) ON (r.writtenAt);
 
+CREATE INDEX compiled_truth_revision_twin IF NOT EXISTS
+FOR (r:CompiledTruthRevision) ON (r.twin);
+
 // ----------------------------------------------------------
 // :Report node — durable, keyword-routed output of admin
 // workflows (daily briefing, dream cycle, ad-hoc analyses).

package/payload/platform/plugins/.claude-plugin/marketplace.json

@@ -99,6 +99,11 @@
       "source": "./scheduling",
       "version": "0.1.0"
     },
+    {
+      "name": "substack-import",
+      "source": "./substack-import",
+      "version": "0.1.0"
+    },
     {
       "name": "telegram",
       "source": "./telegram",

package/payload/platform/plugins/admin/.claude-plugin/plugin.json

@@ -1,6 +1,6 @@
 {
   "name": "admin",
-  "description": "Platform administration plugin. Provides system-status, public-hostname (deterministic Cloudflare public-URL resolver, single call returning the operator's canonical hostname so agents never guess property names on :CloudflareHostname nodes), publish-site (move an extracted static-site tree under <accountDir>/sites/<slug>/, emit one canonical path slug, and refresh llms.txt at the site root — typed refusal taxonomy replaces the prior skill-prose contract), brand-settings, account-manage, account-update, admin-add, admin-remove, admin-list, admin-update-pin, agent-list, agent-config-read, logs-read, plugin-read, skill-load (one-call resolve+read for SKILL.md by skill name, the canonical primitive for loading a named skill; plugin-read remains the reader for references/* and PLUGIN.md), store-skill (deterministic write counterpart to plugin-read; persists operator-authored skills as plugin files under the active account), session-reset, session-resume, file-attach, wifi, action-approval tools (action-pending, action-approve, action-reject, action-edit), and session-retrospective-mark-complete (the deterministic sentinel the session-end Stop hook reads via JSONL grep to release the retrospective gate; admin calls it once per session after running the three-pass retrospective) for managing the Maxy platform.",
+  "description": "Platform administration plugin. Provides system-status, public-hostname (deterministic Cloudflare public-URL resolver, single call returning the operator's canonical hostname so agents never guess property names on :CloudflareHostname nodes), publish-site (move an extracted static-site tree under <accountDir>/sites/<slug>/, emit one canonical path slug, and refresh llms.txt at the site root — typed refusal taxonomy replaces the prior skill-prose contract), brand-settings, account-manage, account-update, admin-add, admin-remove, admin-list, admin-update-pin, agent-list, agent-config-read, logs-read, plugin-read, skill-load (one-call resolve+read for SKILL.md by skill name, the canonical primitive for loading a named skill; plugin-read remains the reader for references/* and PLUGIN.md), store-skill (deterministic write counterpart to plugin-read; persists operator-authored skills as plugin files under the active account), session-reset, session-resume, file-attach, wifi, action-approval tools (action-pending, action-approve, action-reject, action-edit), session-retrospective-mark-complete (the deterministic sentinel the session-end Stop hook reads via JSONL grep to release the retrospective gate; admin calls it once per session after running the three-pass retrospective), and pre-turn-graph-pass-skip (the deterministic decline sentinel the per-turn graph-pass Stop hook reads via JSONL grep to release the per-turn gate when the operator turn has no new writes; admin calls it instead of dispatching to database-operator) for managing the Maxy platform.",
   "version": "0.1.0",
   "author": {
     "name": "Rubytech LLC"

package/payload/platform/plugins/admin/PLUGIN.md

@@ -1,7 +1,7 @@
 ---
 name: admin
 surface: platform
-description: "Platform administration plugin. Provides system-status, public-hostname (deterministic Cloudflare public-URL resolver, single call returning the operator's canonical hostname so agents never guess property names on :CloudflareHostname nodes), publish-site (move an extracted static-site tree under <accountDir>/sites/<slug>/, emit one canonical path slug, and refresh llms.txt at the site root — typed refusal taxonomy replaces the prior skill-prose contract), brand-settings, account-manage, account-update, admin-add, admin-remove, admin-list, admin-update-pin, agent-list, agent-config-read, logs-read, plugin-read, skill-load (one-call resolve+read for SKILL.md by skill name, the canonical primitive for loading a named skill; plugin-read remains the reader for references/* and PLUGIN.md), store-skill (deterministic write counterpart to plugin-read; persists operator-authored skills as plugin files under the active account), session-reset, session-resume, file-attach, wifi, action-approval tools (action-pending, action-approve, action-reject, action-edit), and session-retrospective-mark-complete (the deterministic sentinel the session-end Stop hook reads via JSONL grep to release the retrospective gate; admin calls it once per session after running the three-pass retrospective) for managing the Maxy platform."
+description: "Platform administration plugin. Provides system-status, public-hostname (deterministic Cloudflare public-URL resolver, single call returning the operator's canonical hostname so agents never guess property names on :CloudflareHostname nodes), publish-site (move an extracted static-site tree under <accountDir>/sites/<slug>/, emit one canonical path slug, and refresh llms.txt at the site root — typed refusal taxonomy replaces the prior skill-prose contract), brand-settings, account-manage, account-update, admin-add, admin-remove, admin-list, admin-update-pin, agent-list, agent-config-read, logs-read, plugin-read, skill-load (one-call resolve+read for SKILL.md by skill name, the canonical primitive for loading a named skill; plugin-read remains the reader for references/* and PLUGIN.md), store-skill (deterministic write counterpart to plugin-read; persists operator-authored skills as plugin files under the active account), session-reset, session-resume, file-attach, wifi, action-approval tools (action-pending, action-approve, action-reject, action-edit), session-retrospective-mark-complete (the deterministic sentinel the session-end Stop hook reads via JSONL grep to release the retrospective gate; admin calls it once per session after running the three-pass retrospective), and pre-turn-graph-pass-skip (the deterministic decline sentinel the per-turn graph-pass Stop hook reads via JSONL grep to release the per-turn gate when the operator turn has no new writes; admin calls it instead of dispatching to database-operator) for managing the Maxy platform."
 tools:
   - name: system-status
     publicAllowlist: false
@@ -75,6 +75,9 @@ tools:
   - name: session-retrospective-mark-complete
     publicAllowlist: false
     adminAllowlist: true
+  - name: pre-turn-graph-pass-skip
+    publicAllowlist: false
+    adminAllowlist: true
 hidden:
   - remote-auth-status
   - remote-auth-set-password
@@ -110,7 +113,7 @@ Tools are available via the `admin` MCP server.
 
 `logs-read { type: "agent-stream" }` is the canonical name for the per-conversation tool-use/tool-result archive previously called `system`; both names work and the legacy alias is preserved.
 
-**Stream log is the primary diagnostic surface for an in-progress session.** Every stream log is named by sessionKey and the stream-log file exists from the first token. The single-writer mandate (2026-05-14) mechanically enforces this contract: the single writer module at [`platform/ui/app/lib/claude-agent/stream-log-writer.ts`](../../ui/app/lib/claude-agent/stream-log-writer.ts) owns every open of `claude-agent-stream-*.log`, opens the file lazily on the first SDK token byte (via `streamLog.writeToken`), and buffers pre-token markers in memory while duplicating them to `server.log` via `console.error` so zero-token sessions retain observability. The build gate `platform/ui/scripts/check-stream-log-writer.mjs` rejects any external `appendFileSync`/`createWriteStream` against the `claude-agent-stream-*` pattern outside the writer module, plus any `claude-agent-stream-${conversationId}` literal (the pre-1013 wrong-identifier signature that produced two files for one session on the Beacons Pi 2026-05-14). The first-token invariant is bound by `platform/scripts/__tests__/first-token-creates-stream-log.test.sh`: one operator turn, one token, `claude-agent-stream-<sessionKey>.log` exists and contains the token bytes after the byte returns to the operator. The parent-process console fan-out tee in [`platform/ui/app/lib/claude-agent/logging.ts`](../../ui/app/lib/claude-agent/logging.ts) appends every `[<tag>]`-prefixed `console.error` / `console.log` line to every active session's stream log alongside `server.log`. For diagnosing an in-session issue (WhatsApp inbound, Cloudflare action, persist write, baileys error), call `logs-read { sessionKey: "<…>" }` first — the stream log carries both the agent lifecycle AND the parent-process events that occurred during the session window. The `conversationId` parameter remains a legacy alias resolving to the same sessionKey-named file. `logs-read { type: "server" }` is the cross-session escape hatch for events outside any single session window.
+**Stream log is the primary diagnostic surface for an in-progress session.** Every stream log is named by sessionKey and the stream-log file exists from the first token. The single-writer mandate (2026-05-14) mechanically enforces this contract: the single writer module at [`platform/ui/app/lib/claude-agent/stream-log-writer.ts`](../../ui/app/lib/claude-agent/stream-log-writer.ts) owns every open of `claude-agent-stream-*.log`, opens the file lazily on the first SDK token byte (via `streamLog.writeToken`), and buffers pre-token markers in memory while duplicating them to `server.log` via `console.error` so zero-token sessions retain observability. The build gate `platform/ui/scripts/check-stream-log-writer.mjs` rejects any external `appendFileSync`/`createWriteStream` against the `claude-agent-stream-*` pattern outside the writer module, plus any `claude-agent-stream-${sessionId}` literal (the pre-1013 wrong-identifier signature that produced two files for one session on the Beacons Pi 2026-05-14). The first-token invariant is bound by `platform/scripts/__tests__/first-token-creates-stream-log.test.sh`: one operator turn, one token, `claude-agent-stream-<sessionKey>.log` exists and contains the token bytes after the byte returns to the operator. The parent-process console fan-out tee in [`platform/ui/app/lib/claude-agent/logging.ts`](../../ui/app/lib/claude-agent/logging.ts) appends every `[<tag>]`-prefixed `console.error` / `console.log` line to every active session's stream log alongside `server.log`. For diagnosing an in-session issue (WhatsApp inbound, Cloudflare action, persist write, baileys error), call `logs-read { sessionKey: "<…>" }` first — the stream log carries both the agent lifecycle AND the parent-process events that occurred during the session window. The `sessionId` parameter remains a legacy alias resolving to the same sessionKey-named file. `logs-read { type: "server" }` is the cross-session escape hatch for events outside any single session window.
 
 ## Skills
 
@@ -139,8 +142,9 @@ Tools are available via the `admin` MCP server.
 - `hooks/webfetch-preflight.mjs` — short-circuits WebFetch on JS-SPA shells with a structured `WEBFETCH_CANNOT_READ_JS_SPA` error so the agent surfaces a loud failure to the owner instead of paying the 60s extraction timeout. Fail-open on any internal error.
 - `hooks/askuserquestion-investigate-gate.sh` — PreToolUse matcher=`AskUserQuestion`. Blocks the question (exit 2) when no read-only investigation tool has fired since the latest real user turn in the session JSONL. The structural fix for the failure class where the agent fabricates a menu before evidence-gathering (session `c085ec2c-46fb-4b73-8865-68cf85866ea8` 2026-05-22 — "change remote access password" → invented options "Admin PIN / Cloudflare tunnel / WiFi password" with zero prior tool_use; post-correction the agent immediately fired `remote-auth-status` → `ToolSearch` → `remote-auth-set-password`, proving it knew the moves). **Allowlist** (exact, with trailing `__<tool>` suffix-match for namespaced `mcp__plugin_<plugin>_<server>__<tool>` aliases): `ToolSearch`, `Grep`, `Glob`, `Read`, `LS`, `NotebookRead`, `Bash`, `WebFetch`, `WebSearch`, plus the read-only admin / memory MCP tools (`*-status`, `*-list`, `*-read`, `skill-find`, `memory-find-candidates`, `profile-read`, `conversation-list`, `memory-list-attachments`, `memory-read-attachment`). **Block message:** `Blocked: AskUserQuestion requires at least one investigation tool (ToolSearch, Grep, Read, *-list, *-status, *-read, skill-find, ...) earlier in this turn. Search the operator's literal phrase first.` **Log line** (stderr, one per call): `[ask-gate] decision=<allow|block> sessionId=<id8> seen=<csv|-> reason=<allowlist-hit|no-investigation|fail-open-no-transcript|fail-open-parse-error>`. **Fail-open** on missing transcript or parse error — nudges, never bricks the UI.
 - `hooks/session-end-retrospective.sh` — **Stop hook.** Fires on every assistant end_turn but exits 0 with `trigger-skipped reason=<role-not-admin|is-specialist|empty-stdin|missing-transcript|no-intent-match>` on every path EXCEPT when the operator's latest real-user message carries an end-intent token (`/end`, `/archive`, `end session`, `archive this session`, case-insensitive, must match the whole message or a standalone line — not embedded in prose). On that path the hook blocks (exit 2 + structured retrospective instruction on stderr per Stop-hook contract) and emits `[session-retrospective] gate-blocked sessionId=<id> reason=sentinel-absent` until the admin agent calls the `session-retrospective-mark-complete` MCP sentinel tool. Completion is recognised by greping the operator's own JSONL (`transcript_path` on the Stop envelope) for a `tool_use` block whose `name` equals that exact string — never by parsing prose ([[feedback_no_stdout_parsing_for_control_flow]], [[feedback_doctrine_paragraph_is_not_a_gate]]). The sentinel-grep doubles as the re-entry guard: every Stop after the sentinel call sees it and exits 0 with `gate-released sessionId=<id>`. Recursion guard: any session whose `MAXY_SPECIALIST` env is set (specialist subagent dispatched via Task tool) skips the gate. The Sidebar archive button POSTs `/:sessionId/archive` directly to the manager which kills the PTY — no Stop fires after that, so clicking archive is the operator's current escape hatch (opt-out remains deferred per the task spec). All emits go through `POST /api/admin/log-ingest`; the only stderr writer is the gate-blocked path (the instruction block the agent reads). The retrospective itself runs as one or more additional turns inside the operator's existing admin session — Task-tool delegation to `database-operator` is the existing IDENTITY.md "Recording" route, not a parallel admin session ([[feedback_deterministic_means_remove_llm]]).
-- `hooks/pre-turn-graph-pass.sh` — **UserPromptSubmit hook.** Tasks 425, 430 — supersedes Task 412's Stop-hook registration of the same logic. Task 412 fired on every admin-agent `end_turn`, including the agent's own ack after the dispatch, producing an unbounded loop until context exhaustion. Firing on `UserPromptSubmit` fires once per operator prompt and removes the loop primitive entirely. **Task 430 reshaped the orchestration:** admin is the session-walker. The hook no longer packages the conversation into the specialist's prompt — that collapsed the data channel and instruction channel into one context, so operator prose pattern-matching as a writer-binding instruction ("memory-update is denied this session") was treated as binding and the pass self-denied (session `d7070b66-1759-41ca-a771-d7559b8a1967` retrospective wrote 3 graph rows on the same session where the pre-turn pass returned `writes=0`). The Task 425 → Task 430 shape change brings the per-turn pass into line with the session-end retrospective, which never had this defect: admin walks, admin dispatches per-write briefs to `database-operator`. The hook now `GET /api/admin/post-turn-context?conversationId&accountId` to fetch every node already written under this conversationId, then emits the documented UserPromptSubmit stdout JSON envelope `{"hookSpecificOutput":{"hookEventName":"UserPromptSubmit","additionalContext":"<directive>"}}` with `exit 0`. The directive body begins with the literal token `[system: pre-turn-graph-pass-hook]` and carries one interpolated `<prior-writes>` block (the skip-list); the IDENTITY.md UserPromptSubmit-directive section routes the agent into scanning the just-submitted operator turn (already in admin's context) and dispatching one `Task → database-operator` per candidate write before composing its reply. Coverage trade: the assistant response to the most recent operator prompt is not covered by a per-turn pass until the next operator turn (or never, if no next turn arrives); end-of-session coverage is provided by `session-end-retrospective.sh` which walks the full transcript at `/end`. Duplicate writes between per-turn dispatches and the admin agent's discretionary inline `database-operator` dispatch are accepted at write time; cleanup is delegated to the dream-cycle / hygiene sweeps (Tasks 410, 411). Early exits: `reason=role-not-admin` (the hook fires only on admin-agent prompt submissions by doctrine), `reason=is-specialist` (`MAXY_SPECIALIST` env-stamped on every subagent PTY blocks the hook from firing inside a specialist's own session), `reason=missing-transcript` (safety exit), `reason=wrong-hook-event event=<…>` (stale Stop-shaped fire), `reason=missing-env env=MAXY_UI_INTERNAL_PORT` (UI port unset). All emits go through `POST /api/admin/log-ingest` under tag `pre-turn-graph-pass`. The canonical fire line is `dispatch sessionId=<id> conversationId=<id> directive-shape=admin-walks priorWritesCount=<n> contextBytes=<n> ms=<n>`; the `directive-shape=admin-walks` field distinguishes the Task 430 shape from any cached device still running the Task 425 shape. Success: one or more `db-op: writes=<n>` lines in the admin session transcript reflecting per-write dispatches admin actually emitted (zero candidate writes ⇒ zero dispatches, no `db-op` line). Failure: a `db-op: error reason=<…>` line, or a turn that named new entities / commitments where admin emitted no dispatch — the latter is the regression signal for self-denial.
-- `hooks/turn-completed-graph-write.sh` — **Dormant since Task 214** — the Stop-hook registration is no longer written to account settings.json; admin delegates graph writes to `database-operator` via the Task tool. The script, envelope walker, loopback `/api/admin/claude-sessions` 127.0.0.1 bypass, `[turn-recorder]` emitters, recorder-auto-archive subscriber, and `initialMessage` envelope spec are preserved as reusable infrastructure for any future autonomous post-turn flow. Historical contract (still describes the dormant path): Stop hook fired once per completed admin-agent turn. Gates on `MAXY_SESSION_ROLE=admin` + `MAXY_SPECIALIST!=database-operator` so it never recurses into the recorder PTY or fires on public sessions. **Task 147** — the recorder is spawned via the same route Sidebar uses. ONE POST to `POST /api/admin/claude-sessions`, body carries `{specialist:'database-operator', model:'haiku', initialMessage:<json-envelope>, adminSessionId:<op>}` — no synthetic `senderId: 'turn-recorder'` marker. The Hono wrapper bypasses cookie auth on this exact method+path when the request originates from `127.0.0.1` (same trust boundary the claude-session-manager itself relies on), looks up the operator's senderId from the manager's `/<adminSessionId>/meta`, and forwards a Sidebar-shape spawn body. The `/recorder-spawn` sibling route is gone. The hook reads its UI port from `MAXY_UI_INTERNAL_PORT` (stamped on the manager systemd unit) — no fallback; absence emits `[turn-recorder] spawn-failed reason=missing-env env=MAXY_UI_INTERNAL_PORT` to stderr instead of silently 19199-ing. **Task 177** — `initialMessage` is a JSON-stringified envelope; the schema lives in [`platform/plugins/docs/references/admin-session.md`](../docs/references/admin-session.md) under "`initialMessage` JSON envelope (Task 177)". Top-level keys exactly: `turns`, `conversationId`, `accountId`, `occurredAt`. `turns` is the full conversation transcript, oldest first, newest last, with each entry `{ role: "user"|"assistant", text, ts, toolCalls? }`. No windowing, no truncation. Multi-record assistant messages collapse on `message.id`. `tool_use` and matching `tool_result` blocks attach as a single `toolCalls` entry on the owning assistant turn; the user record carrying only the `tool_result` does not create a separate user turn. No leading instruction prose. `toolCalls[].input` and `toolCalls[].output` are native JSON values, never re-stringified. (Replaces Task 175's `(operatorMessage, assistantReply)` pair, which asserted a temporal Q→A relationship the walker never enforced.) One observability line `[turn-recorder] envelope sessionId=<op> turnsCount=<n> userTurns=<n> assistantTurns=<n> toolCallTurns=<n>` precedes `spawn-request`. The envelope rides on the `/spawn` body as a trailing positional argv to `claude`, so the database-operator session's JSONL first `role=user` line is the JSON object verbatim. No separate `POST /:id/input` call, no bracketed-paste, no keystroke injection. The recorder-auto-archive subscriber stops the recorder PTY as soon as its JSONL contains an assistant message with `stop_reason === "end_turn"`. **Task 129** — every emit goes through `POST /api/admin/log-ingest` so the lines land in `server.log` keyed by the operator session id. The chain is `trigger` → `spawn-request` → `spawn-success` → `input-delivered` → `tool-call` × N → `tool-result` × N → `write-complete` → `auto-archive`; each gated-off path emits one `trigger-skipped reason=<role-not-admin|is-recorder|empty-stdin|missing-transcript|conversation-empty>` line. Failure modes — `spawn-failed`, `tool-surface-missing`, `input-failed`, `write-empty`, `auto-archive reason=stale-recorder` — each emit one named line; absence is itself a defect.
+- `hooks/per-turn-graph-pass-gate.sh` — **Stop hook.** Task 440. Per-turn blocking gate that closes the gap left when Task 425 moved the per-turn graph pass off Stop onto `UserPromptSubmit`. `pre-turn-graph-pass.sh` delivers the directive (with prior-writes skip-list) as `additionalContext` and exits 0 — that channel can't carry a release path because UserPromptSubmit's only blocking shape erases the prompt without feeding stderr to Claude ([[feedback_doctrine_paragraph_is_not_a_gate]]). This Stop hook is the enforcement primitive: walks the operator's JSONL since the latest real-user message, releases on either a `Task` `tool_use` whose `input.subagent_type` is `database-operator` (admin dispatched the writes) or a `tool_use` whose `name` is `mcp__admin__pre-turn-graph-pass-skip` (admin declined via the deterministic skip sentinel). Both releases emit `gate-released sessionId=<id> via=<dispatch|skip-sentinel>` and exit 0. Absent both, exits 2 with a stderr instruction naming the two release paths — admin reads the instruction (Stop's exit-2 stderr-to-Claude contract) and acts. Recursion guard: any session with `MAXY_SPECIALIST` set (specialist subagent dispatched via Task) skips. Loop-safety: once either release tool_use sits in the JSONL, every subsequent Stop on the same operator-prompt window sees it and exits 0 — the Task 412 loop is avoided because the release decision is JSONL-based, never a re-fire of the directive. Composes with `session-end-retrospective.sh` (the other Stop-array entry): the session-end hook gates only on operator end-intent, this one gates every turn. Early exits: `reason=role-not-admin`, `reason=is-specialist`, `reason=empty-stdin`, `reason=missing-transcript`, `reason=no-real-user` (transcript has no real-user record yet — e.g. SessionStart bootstrap, or only `tool_result`-shaped user records). All emits go through `POST /api/admin/log-ingest` under tag `per-turn-graph-pass-gate`.
+- `hooks/pre-turn-graph-pass.sh` — **UserPromptSubmit hook.** Tasks 425, 430 — supersedes Task 412's Stop-hook registration of the same logic. Task 412 fired on every admin-agent `end_turn`, including the agent's own ack after the dispatch, producing an unbounded loop until context exhaustion. Firing on `UserPromptSubmit` fires once per operator prompt and removes the loop primitive entirely. **Task 430 reshaped the orchestration:** admin is the session-walker. The hook no longer packages the conversation into the specialist's prompt — that collapsed the data channel and instruction channel into one context, so operator prose pattern-matching as a writer-binding instruction ("memory-update is denied this session") was treated as binding and the pass self-denied (session `d7070b66-1759-41ca-a771-d7559b8a1967` retrospective wrote 3 graph rows on the same session where the pre-turn pass returned `writes=0`). The Task 425 → Task 430 shape change brings the per-turn pass into line with the session-end retrospective, which never had this defect: admin walks, admin dispatches per-write briefs to `database-operator`. The hook now `GET /api/admin/post-turn-context?sessionId&accountId` to fetch every node already written under this sessionId, then emits the documented UserPromptSubmit stdout JSON envelope `{"hookSpecificOutput":{"hookEventName":"UserPromptSubmit","additionalContext":"<directive>"}}` with `exit 0`. The directive body begins with the literal token `[system: pre-turn-graph-pass-hook]` and carries one interpolated `<prior-writes>` block (the skip-list); the IDENTITY.md UserPromptSubmit-directive section routes the agent into scanning the just-submitted operator turn (already in admin's context) and dispatching one `Task → database-operator` per candidate write before composing its reply. Enforcement is the Stop-array sibling `per-turn-graph-pass-gate.sh` (Task 440): every admin turn end blocks until admin either dispatched the writes (`Task` `tool_use` with `input.subagent_type=database-operator`) or called the skip sentinel (`pre-turn-graph-pass-skip` MCP tool with a `reason` arg) since the latest real-user message. Duplicate writes between per-turn dispatches and the admin agent's discretionary inline `database-operator` dispatch are accepted at write time; cleanup is delegated to the dream-cycle / hygiene sweeps (Tasks 410, 411). Early exits: `reason=role-not-admin` (the hook fires only on admin-agent prompt submissions by doctrine), `reason=is-specialist` (`MAXY_SPECIALIST` env-stamped on every subagent PTY blocks the hook from firing inside a specialist's own session), `reason=missing-transcript` (safety exit), `reason=wrong-hook-event event=<…>` (stale Stop-shaped fire), `reason=missing-env env=MAXY_UI_INTERNAL_PORT` (UI port unset). All emits go through `POST /api/admin/log-ingest` under tag `pre-turn-graph-pass`. The canonical fire line is `dispatch sessionId=<id> sessionId=<id> directive-shape=admin-walks priorWritesCount=<n> contextBytes=<n> ms=<n>`; the `directive-shape=admin-walks` field distinguishes the Task 430 shape from any cached device still running the Task 425 shape. Success: one or more `db-op: writes=<n>` lines in the admin session transcript reflecting per-write dispatches admin actually emitted (zero candidate writes ⇒ zero dispatches, no `db-op` line). Failure: a `db-op: error reason=<…>` line, or a turn that named new entities / commitments where admin emitted no dispatch — the latter is the regression signal for self-denial.
+- `hooks/turn-completed-graph-write.sh` — **Dormant since Task 214** — the Stop-hook registration is no longer written to account settings.json; admin delegates graph writes to `database-operator` via the Task tool. The script, envelope walker, loopback `/api/admin/claude-sessions` 127.0.0.1 bypass, `[turn-recorder]` emitters, recorder-auto-archive subscriber, and `initialMessage` envelope spec are preserved as reusable infrastructure for any future autonomous post-turn flow. Historical contract (still describes the dormant path): Stop hook fired once per completed admin-agent turn. Gates on `MAXY_SESSION_ROLE=admin` + `MAXY_SPECIALIST!=database-operator` so it never recurses into the recorder PTY or fires on public sessions. **Task 147** — the recorder is spawned via the same route Sidebar uses. ONE POST to `POST /api/admin/claude-sessions`, body carries `{specialist:'database-operator', model:'haiku', initialMessage:<json-envelope>, adminSessionId:<op>}` — no synthetic `senderId: 'turn-recorder'` marker. The Hono wrapper bypasses cookie auth on this exact method+path when the request originates from `127.0.0.1` (same trust boundary the claude-session-manager itself relies on), looks up the operator's senderId from the manager's `/<adminSessionId>/meta`, and forwards a Sidebar-shape spawn body. The `/recorder-spawn` sibling route is gone. The hook reads its UI port from `MAXY_UI_INTERNAL_PORT` (stamped on the manager systemd unit) — no fallback; absence emits `[turn-recorder] spawn-failed reason=missing-env env=MAXY_UI_INTERNAL_PORT` to stderr instead of silently 19199-ing. **Task 177** — `initialMessage` is a JSON-stringified envelope; the schema lives in [`platform/plugins/docs/references/admin-session.md`](../docs/references/admin-session.md) under "`initialMessage` JSON envelope (Task 177)". Top-level keys exactly: `turns`, `sessionId`, `accountId`, `occurredAt`. `turns` is the full conversation transcript, oldest first, newest last, with each entry `{ role: "user"|"assistant", text, ts, toolCalls? }`. No windowing, no truncation. Multi-record assistant messages collapse on `message.id`. `tool_use` and matching `tool_result` blocks attach as a single `toolCalls` entry on the owning assistant turn; the user record carrying only the `tool_result` does not create a separate user turn. No leading instruction prose. `toolCalls[].input` and `toolCalls[].output` are native JSON values, never re-stringified. (Replaces Task 175's `(operatorMessage, assistantReply)` pair, which asserted a temporal Q→A relationship the walker never enforced.) One observability line `[turn-recorder] envelope sessionId=<op> turnsCount=<n> userTurns=<n> assistantTurns=<n> toolCallTurns=<n>` precedes `spawn-request`. The envelope rides on the `/spawn` body as a trailing positional argv to `claude`, so the database-operator session's JSONL first `role=user` line is the JSON object verbatim. No separate `POST /:id/input` call, no bracketed-paste, no keystroke injection. The recorder-auto-archive subscriber stops the recorder PTY as soon as its JSONL contains an assistant message with `stop_reason === "end_turn"`. **Task 129** — every emit goes through `POST /api/admin/log-ingest` so the lines land in `server.log` keyed by the operator session id. The chain is `trigger` → `spawn-request` → `spawn-success` → `input-delivered` → `tool-call` × N → `tool-result` × N → `write-complete` → `auto-archive`; each gated-off path emits one `trigger-skipped reason=<role-not-admin|is-recorder|empty-stdin|missing-transcript|conversation-empty>` line. Failure modes — `spawn-failed`, `tool-surface-missing`, `input-failed`, `write-empty`, `auto-archive reason=stale-recorder` — each emit one named line; absence is itself a defect.
 
 ## Session identifiers (Task 135)
 

package/payload/platform/plugins/admin/hooks/__tests__/archive-ingest-surface-gate.test.sh

@@ -103,7 +103,7 @@ run_case "BYPASS: nested rows[0].archiveType=linkedin + top-level archiveType=wh
   2
 
 run_case "BYPASS: conversation.archiveType=linkedin + top-level archiveType=whatsapp-export → BLOCKED" \
-  '{"hook_event_name":"PreToolUse","tool_name":"mcp__memory__memory-archive-write","tool_input":{"conversation":{"archiveType":"linkedin-connections","conversationId":"x"},"archiveType":"whatsapp-export","ownerNodeId":"x","accountId":"a","rows":[]}}' \
+  '{"hook_event_name":"PreToolUse","tool_name":"mcp__memory__memory-archive-write","tool_input":{"conversation":{"archiveType":"linkedin-connections","sessionId":"x"},"archiveType":"whatsapp-export","ownerNodeId":"x","accountId":"a","rows":[]}}' \
   2
 
 # Plugin-source-edit path block must read tool_input.file_path top-level,