@event4u/agent-config 17.0.0 → 17.2.0

package/README.md

@@ -2,11 +2,12 @@
 
 # Agent Config — every claim machine-checked, including the artifact counts in these badges
 
-[![Smoke](https://github.com/event4u-app/agent-config/actions/workflows/smoke.yml/badge.svg)](https://github.com/event4u-app/agent-config/actions/workflows/smoke.yml) [![Public install smoke (3 OS × 2 Node)](https://github.com/event4u-app/agent-config/actions/workflows/smoke-public-install.yml/badge.svg)](https://github.com/event4u-app/agent-config/actions/workflows/smoke-public-install.yml) [![npm](https://img.shields.io/npm/v/@event4u/agent-config?style=flat-square&label=npm&color=orange)](https://www.npmjs.com/package/@event4u/agent-config) [![agent-config MCP server](https://glama.ai/mcp/servers/event4u-app/agent-config/badges/score.svg)](https://glama.ai/mcp/servers/event4u-app/agent-config) [![MCP Toplist](https://mcptoplist.com/badge/glama%2Fevent4u-app%2Fagent-config.svg)](https://mcptoplist.com/server/glama%2Fevent4u-app%2Fagent-config) [![downloads](https://img.shields.io/npm/dm/@event4u/agent-config?style=flat-square&label=downloads%2Fmonth&color=orange)](https://www.npmjs.com/package/@event4u/agent-config)
+[![Smoke](https://github.com/event4u-app/agent-config/actions/workflows/smoke.yml/badge.svg)](https://github.com/event4u-app/agent-config/actions/workflows/smoke.yml) [![Public install smoke (3 OS × 2 Node)](https://github.com/event4u-app/agent-config/actions/workflows/smoke-public-install.yml/badge.svg)](https://github.com/event4u-app/agent-config/actions/workflows/smoke-public-install.yml) [![npm](https://img.shields.io/npm/v/@event4u/agent-config?style=flat-square&label=npm&color=orange)](https://www.npmjs.com/package/@event4u/agent-config) [![agent-config MCP server](https://glama.ai/mcp/servers/event4u-app/agent-config/badges/score.svg)](https://glama.ai/mcp/servers/event4u-app/agent-config) [![MCP Toplist](https://mcptoplist.com/badge/glama%2Fevent4u-app%2Fagent-config.svg)](https://mcptoplist.com/server/glama%2Fevent4u-app%2Fagent-config)
 
-[![Skills](https://img.shields.io/badge/Skills-301-orange?style=flat-square)](dist/agent-src/skills/) [![Rules](https://img.shields.io/badge/Rules-122-orange?style=flat-square)](dist/agent-src/rules/) [![Commands](https://img.shields.io/badge/Commands-209-orange?style=flat-square)](dist/agent-src/commands/) [![Guidelines](https://img.shields.io/badge/Guidelines-121-orange?style=flat-square)](docs/guidelines/) [![Personas](https://img.shields.io/badge/Personas-29-orange?style=flat-square)](dist/agent-src/personas/) [![Advisors](https://img.shields.io/badge/Advisors-5-orange?style=flat-square)](dist/agent-src/personas/advisors/) <!-- claim:published-artifact-counts -->
+[![Skills](https://img.shields.io/badge/Skills-301-orange?style=flat-square)](dist/agent-src/skills/) [![Rules](https://img.shields.io/badge/Rules-122-orange?style=flat-square)](dist/agent-src/rules/) [![Commands](https://img.shields.io/badge/Commands-210-orange?style=flat-square)](dist/agent-src/commands/) [![Guidelines](https://img.shields.io/badge/Guidelines-121-orange?style=flat-square)](docs/guidelines/) [![Personas](https://img.shields.io/badge/Personas-29-orange?style=flat-square)](dist/agent-src/personas/) [![Advisors](https://img.shields.io/badge/Advisors-5-orange?style=flat-square)](dist/agent-src/personas/advisors/) <!-- claim:published-artifact-counts -->
+[![downloads](https://img.shields.io/npm/dm/@event4u/agent-config?style=flat-square&label=downloads%2Fmonth&color=orange)](https://www.npmjs.com/package/@event4u/agent-config)
 
-<sub>**How these are counted** — one canonical counter, `agent-config` → `update_counts --check`, re-derives all six from the tree and fails CI when one is off by one. Two bases are not what the linked directory shows, so they are stated here rather than left to be inferred: **Commands 209** counts every command file **recursively** (the linked directory holds 62 at its top level) and **Rules 122** counts the **source** rules while the linked projection holds 121 — one rule is dormant and is not projected. **Personas 29** excludes the directory README. Counts are of files and directories: none of them measures quality, activation or adoption. <!-- claim:published-artifact-counts --></sub>
+<sub>**How these are counted** — one canonical counter, `agent-config` → `update_counts --check`, re-derives all six from the tree and fails CI when one is off by one. Two bases are not what the linked directory shows, so they are stated here rather than left to be inferred: **Commands 210** counts every command file **recursively** (the linked directory holds 62 at its top level) and **Rules 122** counts the **source** rules while the linked projection holds 121 — one rule is dormant and is not projected. **Personas 29** excludes the directory README. Counts are of files and directories: none of them measures quality, activation or adoption. <!-- claim:published-artifact-counts --></sub>
 
 **Try it in 30 seconds** — drop one read-only subagent into any repo and watch it check "done": `@production-validator check this branch is actually done`. No wizard, no lock-in, nothing else installed — the [30-second wedge ↓](#quickstart) is the whole first step. Start at the proof, not the catalog: **[event4u-app.github.io/agent-config/proof/](https://event4u-app.github.io/agent-config/proof/)**.
 

package/dist/agent-src/commands/project/analyze.md

@@ -31,7 +31,9 @@ Gather in parallel:
 - Check for: `artisan`, `Makefile`, `Taskfile.yml`, `docker-compose.yml`
 - Check for: `phpstan.neon`, `ecs.php`, `rector.php`, `config-dev/`
 - Check for: `.github/workflows/`, `CODEOWNERS`, `.editorconfig`
-- Check for module roots (resolve via `scripts/_lib/agent_settings.ts::enumerate_modules()`; Laravel shape: `app/Modules/`), multi-tenant indicators (`customer_database`)
+- Check for module roots (resolve via `scripts/_lib/agent_settings.ts::enumerate_modules()`; Laravel shape: `app/Modules/`)
+- Probe the database connections the project declares (`config/database.php` `connections`, `DB_CONNECTION*` keys in `.env.example`, `$connection` properties on models); more than one connection used by models is a multi-database / multi-tenant indicator — name the connections as found, never assume their names
+- Build or refresh the code map: `agent-config code-graph map`. Structure (areas, hubs, cycles, request paths) is read from `agents/map/MAP.md` below, not re-derived by hand <!-- ref-ignore -->
 - Read: `AGENTS.md`, `.github/copilot-instructions.md` (if exist)
 
 **Assurance dimensions — do NOT hand-detect these.** Run
@@ -137,6 +139,7 @@ full pass, so "continue" is dominant and asking is a routine confirmation.
 Analyze the directory structure:
 
 - Map top-level directories with file counts
+- Read `agents/map/MAP.md` first: its areas, infrastructure files, cycles and request paths ARE the structure section — link to it rather than restating it <!-- ref-ignore -->
 - Identify pattern: MVC, modules, services, repositories, DTOs
 - Count: Models, Controllers, Services, Jobs, Commands, Events, Policies
 - Detect: multi-tenancy, API versioning, queue usage
@@ -150,13 +153,14 @@ ARCHITECTURE:
 
   Pattern:      {Modular MVC + Service Layer + Repository}
   API:          {Versioned REST (/api/v1/, /api/v2/)}
-  Multi-Tenant: {✅ Dual-DB (api_database + customer_database) | ❌}
+  Databases:    {connections the models use, as probed — e.g. "2 (mysql, tenant)" | 1}
+  Map:          agents/map/MAP.md ({n} areas · largest {name})
 
 ───────────────────────────────────────────────
 CODE INVENTORY:
 ───────────────────────────────────────────────
 
-  Models:        {count}  (api_database: {n}, customer_database: {n})
+  Models:        {count}  (per connection, when more than one is in use)
   Controllers:   {count}
   Services:      {count}
   Repositories:  {count}
@@ -288,9 +292,8 @@ Write all remaining analysis files that haven't been created yet:
 - `agents/evidence/analysis/architecture/database.md` — schema, connections, multi-tenancy
 - `agents/evidence/analysis/architecture/api.md` — versioning, routes, middleware, auth
 - `agents/evidence/analysis/architecture/infrastructure.md` — Docker, CI/CD, deployment
-- `agents/evidence/analysis/architecture/patterns.md` — design patterns used
-- `agents/evidence/analysis/models/api-database.md` — all api_database models
-- `agents/evidence/analysis/models/customer-database.md` — all customer_database models
+- `agents/evidence/analysis/architecture/patterns.md` — design patterns used; for module and area structure, link `agents/map/MAP.md` instead of restating it <!-- ref-ignore -->
+- `agents/evidence/analysis/models/<connection>.md` — one per database connection the probe found, all models on it
 - `agents/evidence/analysis/modules/{module}.md` — one per module
 - `agents/evidence/analysis/testing/test-map.md` — test suites, coverage, strategy
 

package/dist/agent-src/commands/project/map.md

@@ -0,0 +1,58 @@
+---
+model_tier: lite
+name: project-map
+pack: engineering-base
+visibility: internal
+sub: map
+cluster: project
+skills: [code-intelligence]
+description: Draw the code map into agents/map/
+suggestion:
+  eligible: false
+  rationale: "Cluster sub-command — reached via its cluster head's routing or its explicit /cluster:sub name; not independently suggested (surface-consolidation)."
+workspaces:
+  - agent-config-maintainer
+packs:
+  - meta
+---
+
+# project-map
+
+## Instructions
+
+Build the map of the project the user is standing in, then point at it. The map is
+deterministic and local: no model call, no network, nothing committed by default
+(`agents/map/` is gitignored, ADR-297). <!-- ref-ignore -->
+
+### 1. Build
+
+```bash
+agent-config code-graph map
+```
+
+It builds or updates the code graph first when it is absent or behind, then writes
+`agents/map/index.html`, `agents/map/MAP.md`, one `agents/map/areas/<id>.md` per <!-- ref-ignore -->
+area of five or more files, and `agents/map/map.json`. With Playwright present it <!-- ref-ignore -->
+also renders the page at three widths and reports the render check; without it the
+check reads `unverified`, which is not a pass.
+
+`--root <path>` maps another checkout; `--open` opens the page; `--no-verify`
+skips the render check.
+
+### 2. Report
+
+Quote the command's summary lines (files, areas, largest area, graph state, render
+check) and give the path to `agents/map/index.html`. Do not paraphrase the counts. <!-- ref-ignore -->
+
+### 3. Answer from the map, when the user asked a question
+
+Read `agents/map/MAP.md` first, then the page of the area in question. Say which <!-- ref-ignore -->
+answers came from the map and which from a follow-up `code-graph query` or a file
+read — the map groups files by coupling, it does not explain what they do.
+
+## Rules
+
+- **Never edit `agents/map/` by hand.** Every file there is regenerated. <!-- ref-ignore -->
+- **A `behind:N` or `edited` graph state is part of the answer.** Say it.
+- **"Isolated" is not "dead".** A file reached by reflection, a string path or a
+  framework convention the engine does not read is isolated too.

package/dist/agent-src/commands/project.md

@@ -2,10 +2,10 @@
 model_tier: medium
 name: project
 disable-model-invocation: true
-argument-hint: "[analyze|health]"
+argument-hint: "[analyze|health|map]"
 pack: engineering-base
-intent: "Project dispatcher — full analysis or quick health check"
-routes_to: [project-analyze, project-health]
+intent: "Project dispatcher — full analysis, quick health check, or the project map"
+routes_to: [project-analyze, project-health, project-map]
 replaces: []
 visibility: internal
 description: Project orchestrator — routes to analyze (full audit) and health (read-only status check)
@@ -35,6 +35,7 @@ Top-level orchestrator for the `/project` family — project-wide inspection.
 |---|---|---|
 | `/project analyze` | `commands/project/analyze.md` | Full project analysis — detect stack, inventory modules, audit docs, create missing contexts |
 | `/project health` | `commands/project/health.md` | Quick health check — status of docs, modules, contexts, roadmaps; creates nothing |
+| `/project map` | `commands/project/map.md` | Draw the code map into `agents/map/` — areas, files, links; offline, gitignored |
 
 Sub-command names match the locked contract in
 [`docs/contracts/command-clusters.md`](../docs/contracts/command-clusters.md).
@@ -50,6 +51,7 @@ Sub-command names match the locked contract in
 
    > 1. analyze — full project audit (writes missing contexts)
    > 2. health — read-only status check (creates nothing)
+   > 3. map — draw the code map into `agents/map/`
 
 ## Rules
 

package/dist/agent-src/contexts/execution/host-capability-manifest.md

@@ -37,7 +37,9 @@ host, never a configuration decision (always-on orchestration, Phase 1):
 2. A live environment probe merged on top: `agent_teams` resolves `true`
    when `CLAUDE_CODE_EXPERIMENTAL_AGENT_TEAMS` is set in the process
    environment — the one host fact this repo can observe about Claude
-   Code's experimental multi-instance Agent Teams feature.
+   Code's experimental multi-instance Agent Teams feature. On `claude` the
+   same flag also sets `worker_resume`: the by-name re-task was observed
+   there only under it (cold-start 1.4, 2.1.306; council 2026-10-22).
 3. `SAFE_DEFAULT` (all `false`) for an unrecognized host or an unset probe.
 
 `resolveHostCapabilities(hostId, override?)` still accepts an optional
@@ -92,7 +94,7 @@ SAFE DEGRADATION, NOT BECAUSE ANYTHING WAS MEASURED.
 
 The seven booleans look alike and are not. `HOST_CAPABILITY_REGISTRY` holds
 exactly one row today, so on every other host **all six** fields are the safe
-default — and `agent_teams` is the only field any live check touches. The name
+default — and `agent_teams` (plus `worker_resume` on `claude`) is the only field any live check touches. The name
 `probeHostCapabilities` says "detected" about values that were mostly asserted
 at authoring time.
 
@@ -136,7 +138,7 @@ A prober that decides for itself what counts as evidence is the inference
 | `status_polling` | a **completed poll returning a state the parent did not already hold**. A poll that returns what the parent knew is not an observation of polling |
 | `separate_quota_pool` | a **child leg continuing across a parent quota condition** — the parent constrained, the child proceeding |
 | `worker_respawn` | **one task continuing across a killed and re-spawned worker, same task id.** Never inferred from the fact that spawning and killing both exist separately |
-| `worker_resume` | **a finished leg re-tasked by name whose next `SubagentStart` carries the same ledger `ref`, and whose first call in that run reads more than half its input from cache.** Never inferred from the host documenting a resume or message tool |
+| `worker_resume` | **a finished leg re-tasked by name whose next `SubagentStart` carries the same ledger `ref`, and whose first call in that run reads more than half its input from cache.** Never inferred from the host documenting a resume or message tool. On `claude` the agent-teams flag stands in for this row: the by-name re-task was observed only under it (2.1.306), so the live probe sets the field from the flag (council 2026-10-22) |
 | `fork_shares_cache` | **a fork whose first call reads more than half its input from cache and whose tool list equals the parent's.** Never inferred from documentation |
 | `structured_ask` | **the host's delivered tool surface, in a real session, carrying (or not carrying) a question-picker tool.** The delivered surface is the observation — never the vendor's documentation, and never another host by analogy. A `false` written from an observed surface is "checked, absent"; omitting the field is "never looked" |
 | `reads_project_mcp_config` | **a server named only in the project-scope config appearing in that host's own tool list.** Never inferred from the host documenting MCP support, which does not say which scope it reads |

package/dist/agent-src/skills/code-intelligence/SKILL.md

@@ -38,6 +38,10 @@ code-graph and no appetite to build one — plain `grep`/read is right there.
 
 ## Procedure
 
+0. **Orient first, when the question is "where does X live" rather than "who
+   calls X".** If `agents/map/MAP.md` exists, read it: it names the project's <!-- ref-ignore -->
+   areas (files grouped by coupling), its infrastructure files and its cycles.
+   No map yet? `agent-config code-graph map` builds one in seconds. Then query.
 1. **Detect the source.** `agent-config code-graph detect` — is a fresh consumer
    `graph.json` or native cache present? (A consumer-shipped fresh index wins;
    the native engine covers stale-or-absent — ADR-124 § 2.)
@@ -246,6 +250,8 @@ paragraph claiming "the same five answers" directly after a list of four:
 | `graph_query` | `query` |
 | `graph_path` | `path` |
 | `graph_node` | no CLI verb — one symbol's location, in-edges, out-edges and degree |
+| `graph_map_status` | whether `agents/map/` exists, its counts and freshness | <!-- ref-ignore -->
+| `graph_area` | one area of the map: files, hubs, what it depends on |
 
 `untested` has no tool of its own; it is a flag on `graph_impact`, because the
 two answer the same question over the same diff and a separate tool would have

package/dist/agent-src/templates/agents/agent-project-settings.example.yml

@@ -39,7 +39,7 @@ schema_version: 1
 # CI guard: a release bump of `package.json` must update this value
 # in lockstep — see scripts/check_template_pin_drift.ts (road-to-
 # portable-runtime-and-update-check P3.3).
-agent_config_version: "17.0.0"
+agent_config_version: "17.2.0"
 
 # --- Project identity ---
 project:

package/dist/agent-src/templates/roadmaps.md

@@ -731,7 +731,11 @@ that was never a judgement call.
     (ADR-268 § 10). `resolved by` is one of `evidence`, `agent`,
     `independent:<session or model>`, `council:<record>`, `team:<record>`,
     `owner` — the prefixed forms carry their record, because a resolver
-    with no record is a claim rather than a resolution.
+    with no record is a claim rather than a resolution. The three locked
+    classes (`product-owned`, `business-owned`, `destructive-owned`) take
+    `owner` only; a council answer on one is recorded with `owner` and a
+    decision cell starting `proposed`. A resolved human-only blocker needs
+    a dated `owner` row naming it.
 
     `revisit if` names the condition that reopens the row. Execution reads
     the table **before any step**, and a closed decision is re-asked only

package/dist/cli/registry.js

@@ -77,7 +77,7 @@ export const REGISTRY = [
     { name: 'affected', disposition: 'delegate', synopsis: 'Artefacts related to <artefact> via the discovery relation-graph (BFS).' },
     { name: 'graph-explain', disposition: 'delegate', synopsis: 'Seed on a <concept>, expand 2 hops over the discovery relation-graph with a node budget.' },
     { name: 'benchmark', disposition: 'delegate', synopsis: 'Report context-token reduction vs the full always-loaded projection.' },
-    { name: 'code-graph', disposition: 'delegate', synopsis: 'Deterministic code-graph engine: build|detect|refresh|validate|query|affected|path|explain, plus the gate verbs impact|tests-for|untested|dead.' },
+    { name: 'code-graph', disposition: 'delegate', synopsis: 'Deterministic code-graph engine: build|detect|refresh|validate|query|affected|path|explain|map, plus the gate verbs impact|tests-for|untested|dead.' },
     { name: 'hooks:install', disposition: 'delegate', synopsis: 'Install the combined pre-commit hook.' },
     { name: 'hooks:status', disposition: 'delegate', synopsis: 'Print the runtime hook matrix (per-platform install + bindings).' },
     { name: 'hooks:doctor', disposition: 'delegate', synopsis: 'Diagnose hook health: concerns, posture, missing trampolines (read-only).' },

package/dist/cli-delegate/{chunk-IROAQ77A.js → chunk-3PK3VZHB.js}

@@ -1,4 +1,7 @@
 import { createRequire as __acCreateRequire } from 'node:module'; const require = globalThis.require ?? __acCreateRequire(import.meta.url);
+import {
+  getLogPath
+} from "./chunk-W5JXMF2T.js";
 import {
   classifyOwnership
 } from "./chunk-J2SJZICO.js";
@@ -7,14 +10,21 @@ import {
 var SIDECAR_SUFFIX = ".agent-config.new";
 var EXIT_COMPLETED_WITH_CONFLICTS = 3;
 function isLedgerMigration(inputs) {
-  if (!inputs.exists || inputs.force) return false;
+  if (!inputs.exists || inputs.force || inputs.loggedWrite === true) return false;
   if (inputs.ledgerListed !== true) return false;
   if (inputs.onDiskSha256 === null) return false;
   return classifyOwnership(inputs.recordedSha256, inputs.onDiskSha256) === "unknown";
 }
+function logProofApplies(inputs) {
+  if (!inputs.exists || inputs.force || inputs.onDiskSha256 === null) return false;
+  const ownership = classifyOwnership(inputs.recordedSha256, inputs.onDiskSha256);
+  if (ownership === "recorded-modified") return inputs.recordedIsStaged === true;
+  return ownership === "unknown" && isLedgerMigration({ ...inputs, loggedWrite: false });
+}
 function decideDeployWrite(inputs) {
   if (!inputs.exists) return "write";
   if (inputs.force) return "write";
+  if (inputs.loggedWrite === true && logProofApplies(inputs)) return "write";
   const ownership = classifyOwnership(inputs.recordedSha256, inputs.onDiskSha256);
   if (ownership === "recorded-modified") return "preserve";
   if (ownership === "unknown" && isLedgerMigration(inputs)) {
@@ -39,13 +49,97 @@ function foreignSidecarMessage(target, sidecar) {
   return `Refusing to replace ${sidecar}: it already exists with different content and this install cannot show it wrote it. Package content for ${target} was NOT staged. Move or delete that file and re-run, or rerun with --force to replace the managed file instead.`;
 }
 
+// src/install/loggedWrites.ts
+import { gunzipSync } from "node:zlib";
+import { existsSync, readdirSync, readFileSync, realpathSync, statSync } from "node:fs";
+import { basename, dirname, join, relative, resolve, sep } from "node:path";
+var WRITE_MARKER = '"kind":"write"';
+var fileCache = /* @__PURE__ */ new Map();
+var dirRealpaths = /* @__PURE__ */ new Map();
+function canonicalPath(p) {
+  const abs = resolve(p);
+  const dir = dirname(abs);
+  let real = dirRealpaths.get(dir);
+  if (real === void 0) {
+    try {
+      real = realpathSync(dir);
+    } catch {
+      real = dir;
+    }
+    dirRealpaths.set(dir, real);
+  }
+  return join(real, basename(abs));
+}
+function logFiles(logPath) {
+  const dir = dirname(logPath);
+  if (!existsSync(dir)) return [];
+  const stem = basename(logPath).replace(/\.jsonl$/, "");
+  try {
+    return readdirSync(dir).filter((n) => n.startsWith(stem) && (n.endsWith(".jsonl") || n.endsWith(".jsonl.gz"))).map((n) => join(dir, n)).sort();
+  } catch {
+    return [];
+  }
+}
+function writesIn(file) {
+  let stamp;
+  try {
+    const st = statSync(file);
+    stamp = `${st.size}:${Math.trunc(st.mtimeMs)}`;
+  } catch {
+    return [];
+  }
+  const cached = fileCache.get(file);
+  if (cached !== void 0 && cached.stamp === stamp) return cached.entries;
+  let text = "";
+  try {
+    const raw = readFileSync(file);
+    text = (file.endsWith(".gz") ? gunzipSync(raw) : raw).toString("utf8");
+  } catch {
+    text = "";
+  }
+  const entries = [];
+  for (const line of text.split("\n")) {
+    if (!line.includes(WRITE_MARKER)) continue;
+    let entry;
+    try {
+      entry = JSON.parse(line);
+    } catch {
+      continue;
+    }
+    if (typeof entry.path !== "string" || typeof entry.sha256 !== "string" || entry.sha256 === "") continue;
+    entries.push([canonicalPath(entry.path), entry.sha256]);
+  }
+  fileCache.set(file, { stamp, entries });
+  return entries;
+}
+function loggedWriteDigests(logPath = getLogPath()) {
+  const writes = /* @__PURE__ */ new Map();
+  for (const file of logFiles(logPath)) {
+    for (const [target, sha] of writesIn(file)) {
+      let set = writes.get(target);
+      if (set === void 0) writes.set(target, set = /* @__PURE__ */ new Set());
+      set.add(sha);
+    }
+  }
+  return writes;
+}
+function onlyDeployed(anchor, currentFiles, log) {
+  return (target) => {
+    const rel = relative(resolve(anchor), resolve(target)).split(sep).join("/");
+    if (currentFiles.has(rel)) log(target);
+  };
+}
+function wroteTheseBytes(target, onDiskSha256, writes) {
+  return writes.get(canonicalPath(target))?.has(onDiskSha256) === true;
+}
+
 // src/install/plan.ts
 import { createHash } from "node:crypto";
-import { readFileSync, readdirSync, statSync } from "node:fs";
-import { join, relative, resolve } from "node:path";
+import { readFileSync as readFileSync2, readdirSync as readdirSync2, statSync as statSync2 } from "node:fs";
+import { join as join2, relative as relative2, resolve as resolve2 } from "node:path";
 function sha256File(path) {
   try {
-    const data = readFileSync(path);
+    const data = readFileSync2(path);
     return createHash("sha256").update(data).digest("hex");
   } catch {
     return null;
@@ -66,15 +160,15 @@ function walkSourceTree(dir) {
 function walkInto(dir, out) {
   let entries;
   try {
-    entries = readdirSync(dir).sort();
+    entries = readdirSync2(dir).sort();
   } catch {
     return;
   }
   for (const name of entries) {
-    const full = join(dir, name);
+    const full = join2(dir, name);
     let isDir;
     try {
-      isDir = statSync(full).isDirectory();
+      isDir = statSync2(full).isDirectory();
     } catch {
       continue;
     }
@@ -94,8 +188,8 @@ function buildInstallPlan(inputs) {
       files = files.filter(source.fileFilter);
     }
     for (const srcFile of files) {
-      const rel = relative(source.srcDir, srcFile);
-      const destPath = resolve(source.destDir, rel);
+      const rel = relative2(source.srcDir, srcFile);
+      const destPath = resolve2(source.destDir, rel);
       bucket.push(
         fileEntry(srcFile, source.kind, {
           hashContent: source.kind !== "bridge"
@@ -126,12 +220,16 @@ function buildInstallPlan(inputs) {
 export {
   EXIT_COMPLETED_WITH_CONFLICTS,
   isLedgerMigration,
+  logProofApplies,
   decideDeployWrite,
   sidecarPathFor,
   preservedFileMessage,
   migrationPreservedMessage,
   conflictSummaryMessage,
   foreignSidecarMessage,
+  loggedWriteDigests,
+  onlyDeployed,
+  wroteTheseBytes,
   sha256File,
   buildInstallPlan
 };

package/dist/cli-delegate/{chunk-2SFOOFJJ.js → chunk-BZJ4Y3CI.js}

@@ -12,7 +12,7 @@ import {
   EGRESS_RULES,
   evaluateMatch,
   shannonEntropy
-} from "./chunk-E6XNBVE4.js";
+} from "./chunk-KR5ZIFPO.js";
 import {
   SRC_SKILLS
 } from "./chunk-YCU3TIXH.js";

package/dist/cli-delegate/{chunk-CAL76ADI.js → chunk-CMZEKSNN.js}

@@ -42,7 +42,7 @@ import {
   readUserPin,
   residualReport,
   runtimeRootFor
-} from "./chunk-GOMILC3O.js";
+} from "./chunk-QBHEI5N4.js";
 import {
   LEGACY_ALL,
   excludedRuleBasenames,
@@ -89,7 +89,7 @@ import {
   readCanaryStatusLine,
   runCensus,
   summaryLine
-} from "./chunk-2SFOOFJJ.js";
+} from "./chunk-BZJ4Y3CI.js";
 import {
   EGRESS_RULES,
   REVIEW_CREDENTIAL_ROWS,
@@ -101,7 +101,7 @@ import {
   replaceSpans,
   resolveOverlaps,
   vendorPrefix
-} from "./chunk-E6XNBVE4.js";
+} from "./chunk-KR5ZIFPO.js";
 import {
   classifyFileState,
   classifyOwnership,
@@ -468,7 +468,7 @@ function censusRuleDir(dir) {
 }
 
 // src/scripts/cache_realization_report.ts
-import { execFileSync } from "node:child_process";
+import { execFileSync as execFileSync2 } from "node:child_process";
 import * as fs7 from "node:fs";
 import { homedir as homedir2 } from "node:os";
 import * as path12 from "node:path";
@@ -809,6 +809,7 @@ if (process.argv[1] !== void 0) {
 }
 
 // src/scripts/hooks/subagent_ledger_hook.ts
+import { execFileSync } from "node:child_process";
 import * as crypto from "node:crypto";
 import * as fs5 from "node:fs";
 import * as path10 from "node:path";
@@ -1069,6 +1070,41 @@ function appendLedgerLine(root, ts, line) {
   fs5.appendFileSync(file, `${JSON.stringify(line)}
 `, "utf8");
 }
+var INDEX_SUBDIR = "index";
+function writeIndexEntry(root, ts, key, value) {
+  if (is_replay_mode()) return;
+  const dir = path10.join(root, LEDGER_DIR, INDEX_SUBDIR, ts.slice(0, 7));
+  fs5.mkdirSync(dir, { recursive: true });
+  const tmp = path10.join(dir, `.${key}.${process.pid}.tmp`);
+  try {
+    fs5.writeFileSync(tmp, value, "utf8");
+    fs5.renameSync(tmp, path10.join(dir, key));
+  } finally {
+    fs5.rmSync(tmp, { force: true });
+  }
+}
+function headOf(dir) {
+  try {
+    const env = Object.fromEntries(Object.entries(process.env).filter(([k]) => !k.startsWith("GIT_")));
+    return execFileSync("git", ["-C", dir, "rev-parse", "HEAD"], { encoding: "utf8", timeout: 2e3, stdio: ["ignore", "pipe", "ignore"], env }).trim() || null;
+  } catch {
+    return null;
+  }
+}
+function stateStamp(cwd) {
+  let epoch = null;
+  try {
+    epoch = fs5.statSync(fileURLToPath(import.meta.url)).mtimeMs;
+  } catch {
+  }
+  return typeof cwd === "string" && cwd ? { head: headOf(cwd), cwd_key: refFor(cwd), epoch } : { head: null, cwd_key: null, epoch };
+}
+function stopIndexKey(sessionId, ref) {
+  return refFor(`subagent-stop\0${sessionId}\0${ref}`);
+}
+function recordStopIndex(root, ts, sessionId, ref) {
+  writeIndexEntry(root, ts, stopIndexKey(sessionId, ref), "");
+}
 var WALL_CLOCK_ARMS_MS = [
   { label: "5m", ms: 5 * 60 * 1e3 },
   { label: "15m", ms: 15 * 60 * 1e3 },
@@ -1236,8 +1272,13 @@ function handleStop(root, payload, nowIso, sessionId) {
     // Beside the count, never instead of it: the count says how wrong the
     // candidate was, this says whether it was aimed at the envelope.
     envelope_field_hits: parse3.field_hits,
-    concurrent_open: countOpen(root)
+    concurrent_open: countOpen(root),
+    ...stateStamp(payload["cwd"])
   });
+  try {
+    if (sessionId !== null) recordStopIndex(root, nowIso, sessionId, ref);
+  } catch {
+  }
 }
 function processEnvelope(envelope, consumerRoot) {
   try {
@@ -1827,7 +1868,7 @@ function encodeProjectPath(cwd) {
 }
 function listGitWorktrees(cwd) {
   try {
-    const out = execFileSync("git", ["worktree", "list", "--porcelain"], { cwd, encoding: "utf-8", timeout: 1e4 });
+    const out = execFileSync2("git", ["worktree", "list", "--porcelain"], { cwd, encoding: "utf-8", timeout: 1e4 });
     const paths = [];
     for (const line of out.split("\n")) {
       if (line.startsWith("worktree ")) paths.push(line.slice("worktree ".length).trim());
@@ -1900,7 +1941,7 @@ function computeWorktreeFragmentation(projectsRoot, worktreePaths, maxAgeDays) {
 }
 function hostVersion() {
   try {
-    const out = execFileSync("claude", ["--version"], { encoding: "utf-8", timeout: 5e3 }).trim();
+    const out = execFileSync2("claude", ["--version"], { encoding: "utf-8", timeout: 5e3 }).trim();
     return out.length > 0 ? out : "unknown";
   } catch {
     return "unknown";
@@ -8060,9 +8101,10 @@ var CliClient = class extends ExternalAIClient {
   _cli_calls_path;
   /** Attribution label for every booking this client makes. */
   _consumer;
-  /** The session the last answered call ran in; `resumeNext` continues it once, non-blind only. */
+  /** The session the last answered call ran in; `resumeNext` continues it once, never while `blind`. */
   lastSession = null;
   resumeNext = null;
+  blind = false;
   static _AUTH_FAILURE_PATTERNS = [
     "authentication",
     "unauthorized",
@@ -8435,7 +8477,7 @@ var AnthropicCliClient = class extends CliClient {
     void max_tokens;
     return [
       this.binary,
-      ...seatResumeArgs(this.name, this.resumeNext, { blind: false, now_ms: Date.now() }),
+      ...seatResumeArgs(this.name, this.resumeNext, { blind: this.blind, now_ms: Date.now() }),
       "--print",
       // The seat never needs the user environment; only a version the light
       // launch was probed on gets the flag (`LIGHT_LAUNCH_PROBES`).
@@ -8538,7 +8580,7 @@ var OpenAICliClient = class extends CliClient {
     void user_prompt;
     void max_tokens;
     const modelArgs = this.model === OPENAI_CLI_VENDOR_DEFAULT ? [] : ["--model", this.model];
-    const resume = seatResumeArgs(this.name, this.resumeNext, { blind: false, now_ms: Date.now() });
+    const resume = seatResumeArgs(this.name, this.resumeNext, { blind: this.blind, now_ms: Date.now() });
     const bound = resume.length === 0 ? boundFor("openai") : resumeBound(boundFor("openai"));
     return [this.binary, "exec", ...resume, "--json", "--skip-git-repo-check", ...bound, ...modelArgs, "-"];
   }

package/dist/cli-delegate/{chunk-ESUNQ5XR.js → chunk-E3CFULNX.js}

@@ -2,11 +2,15 @@ import { createRequire as __acCreateRequire } from 'node:module'; const require
 import {
   buildInstallPlan,
   isLedgerMigration,
-  sha256File
-} from "./chunk-IROAQ77A.js";
+  logProofApplies,
+  loggedWriteDigests,
+  sha256File,
+  sidecarPathFor,
+  wroteTheseBytes
+} from "./chunk-3PK3VZHB.js";
 import {
   detectToolPresence
-} from "./chunk-A4QJ6ENT.js";
+} from "./chunk-W5JXMF2T.js";
 import {
   classifyOwnership,
   ledgerListedPaths,
@@ -23,10 +27,11 @@ import { existsSync } from "node:fs";
 function isJsonTarget(entry) {
   return entry.kind === "deployed" && entry.path.endsWith(".json");
 }
-function computeConflicts(plan, recordedOverride) {
+function computeConflicts(plan, recordedOverride, loggedOverride) {
   if (plan.policy.force) return [];
   const recorded = recordedOverride ?? recordedHashesWithLedger(plan.root);
   let ledgerFiles = null;
+  let logged;
   const ledgerLists = (p) => {
     ledgerFiles ??= ledgerListedPaths();
     return ledgerFiles.has(p);
@@ -40,6 +45,19 @@ function computeConflicts(plan, recordedOverride) {
       const onDisk = sha256File(entry.path);
       const ownership = classifyOwnership(recorded.get(entry.path), onDisk);
       if (ownership === "recorded-unchanged") continue;
+      const recordedSha = recorded.get(entry.path);
+      const unproven = {
+        exists: true,
+        recordedSha256: recordedSha,
+        onDiskSha256: onDisk,
+        force: false,
+        ledgerListed: ledgerLists(entry.path),
+        recordedIsStaged: typeof recordedSha === "string" && recordedSha === recorded.get(sidecarPathFor(entry.path))
+      };
+      if (logProofApplies(unproven)) {
+        logged ??= loggedOverride ?? loggedWriteDigests();
+        if (wroteTheseBytes(entry.path, onDisk, logged)) continue;
+      }
       if (ownership === "unknown" && plan.policy.knownPaths.has(entry.path)) continue;
       const migration = ownership === "unknown" && isLedgerMigration({
         exists: true,