@mmerterden/multi-agent-pipeline 20.28.0 → 20.30.0

package/CHANGELOG.md

@@ -16,6 +16,25 @@ Internal file-layout changes that don't affect the slash-command surface are sti
 
 ## [Unreleased]
 
+## [20.30.0] - 2026-10-25
+
+### Added
+
+- `/multi-agent:office-on`, `office-off`, `office-status`: a local browser panel over the contract server (127.0.0.1 only). The office is a pixel floor: a lobby whose furniture opens each feature (new run, quest board, PR desk, ledger, autopilot desk with its bot, routines, skills, toolkit, services, commands), one room per repository with one table per run, agents walking in to their chairs, typing or reading by the tool they use, a phase bar over every table, the paper handed on at each phase change, a lamp and a bubble when a run waits for you, finished agents moving to the lounge, and dashed corridors for runs over several repositories. A text list shows the same office. A demo office plays a scripted run when nothing is at work and never sends a write. Text follows `outputLanguage` (English or Turkish). Sprites come from pixel-agents (MIT) with MetroCity characters (CC0), installed in `office/assets/` and served at `/office/assets/` under a fixed file-name pattern; the page is built by `office-build.mjs` from `pipeline/office-src`; attribution in `office/NOTICE.md`. It shows their chats, a cost ledger with the source of every amount, run results, autopilot, routines, stacks, read-only toolkit tools and service reads, and starts runs through the existing launch route. The page installs to `~/.claude/office/` (and the Copilot and Codex trees); `GET /office` serves it, with `MA_OFFICE_PAGE` or `prefs.global.office.page` as overrides. The token reaches the page through a one-time 10 second code, never a URL; the server exits after its idle time and after 8 hours in any case. Remote access is not part of the pipeline; `office-status` reports "not available in this installation" unless a separately installed remote layer is present. Threat model: `multi-agent-refs/features/office-threat-model.md`. The description-surface ceiling rises from 94500 to 95000 bytes for the three commands and their core skills, after their descriptions were trimmed to 812 bytes for all six.
+
+## [20.29.0] - 2026-10-24
+
+### Added
+
+- `run-stop-guard` (Stop hook): a session that drives a pipeline run cannot end its turn mid-run without finishing the current phase, recording an attended wait (`waitingFor: "attended-reply"` or `"local-test"`) or recording a halt. It acts only on the run bound to the session (`sessionId` or the new `hostSessionId` that write-state stamps on every phase move), never on a run matched by worktree alone, lets a turn through when it ends asking the user something, when background tasks or session crons are pending, and after three consecutive holds writes the halt itself. It fails open on any internal error and is switched off with `prefs.global.stopGuard.enabled: false`. SubagentStop is never held.
+- `write-state.mjs` accepts `status: "complete"` only on the run's final phase with Phase 4 evidence (a commit sha, PR or publish, or a noted skip) and a passing plan-coverage verdict; otherwise it exits 6 and writes nothing. Phase 5 reads that exit and reports the run completed only on success; the autopilot runner treats it as `stopped-incomplete` and retries.
+- Attended commit evidence check (`prefs.global.attendedCommitGate.mode`: `warn` by default, `block`, `off`), judged only against a live run.
+- `agent-guard`: AI attribution read from the commit message itself (message, `-F` file, heredoc body, or the whole line when the file is written on that line), skipped hooks (`--no-verify` and its prefixes, `commit -n`, `core.hooksPath`, `HUSKY=0`, `GIT_CONFIG_*`), plain secrets on argv (local development DSNs exempt), and an ask for issue close, Closes/Fixes/Resolves in a PR body and Jira transitions; in bypassPermissions and dontAsk a confirmation record written after the user's answer lets the exact command through once.
+- `run-scope-guard`: Figma MCP calls are refused after Phase 1 of a live, non-analysis run bound to the session; `/multi-agent:analysis` records its run at start. Attended edits outside the run's worktree raise a warning.
+- `prefs.global.readOnlyRepos[]`: checkouts that may not be edited, committed or pushed.
+- `/multi-agent:doctor` reports template hooks that are not registered or not executable and preferences that no registered hook enforces. The installer registers the run hooks itself, records them in `multi-agent-hook-manifest.json` after settings.json is written, and does not re-add a hook the user removed.
+
+
 ## [20.28.0] - 2026-10-24
 
 ### Added

package/README.md

@@ -158,7 +158,7 @@ The discipline behind all of this - bounded loops, evidence gates, token-budgete
 
 ## Commands
 
-`/multi-agent` plus 66 sub-commands. `/multi-agent:help` renders the same catalog in your terminal, in your `outputLanguage`.
+`/multi-agent` plus 69 sub-commands. `/multi-agent:help` renders the same catalog in your terminal, in your `outputLanguage`.
 
 ### Pipeline entries
 
@@ -408,17 +408,17 @@ This enables the matching plugin (+ the shared `ai-common` plugin) in the repo's
 
 ## Tool support
 
-The pipeline runs natively on **Claude Code**, **Copilot CLI** and **Codex CLI** - all three install from the same `pipeline/` source and get the same 66 commands.
+The pipeline runs natively on **Claude Code**, **Copilot CLI** and **Codex CLI** - all three install from the same `pipeline/` source and get the same 69 commands.
 
 | Tool        | Flag                 | What it installs                                                                                       |
 | ----------- | -------------------- | ------------------------------------------------------------------------------------------------------ |
-| Claude Code | `--claude` (default) | slash commands + skills + agents + three `PreToolUse` hooks (secret scan, agent-guard, read-size gate) |
-| Copilot CLI | `--copilot`          | instructions + 66 sub-command skills + scripts                                                         |
-| Codex CLI   | `--codex`            | one router skill + 66 specs as refs + 11 agent TOML + `AGENTS.md` block + `codex mcp add`               |
+| Claude Code | `--claude` (default) | slash commands + skills + agents + the enforcement hooks (secret scan, agent-guard, read-size gate, run-scope guard, run-stop guard) |
+| Copilot CLI | `--copilot`          | instructions + 69 sub-command skills + scripts                                                         |
+| Codex CLI   | `--codex`            | one router skill + 69 specs as refs + 11 agent TOML + `AGENTS.md` block + `codex mcp add`               |
 
 Filter skills by stack with `--platform=ios\|android\|all`.
 
-**Why Codex gets one skill and not 66.** Codex assembles every discovered skill's name
+**Why Codex gets one skill and not 69.** Codex assembles every discovered skill's name
 and description into a single prompt block and drops entries when it overflows, with no
 error. Measured on 0.145: installing one plugin that declares 142 skills surfaced only
 75 of them and evicted an unrelated user skill. So on Codex the pipeline ships a single
@@ -450,6 +450,19 @@ the one host whose panel spans two vendors, and the triage note says so.
 
 The **secret scan** runs as a `PreToolUse` hook on Claude Code (hard-blocks a commit on a hit) and as a pre-push check elsewhere.
 
+**Hooks per host.** The enforcement hooks are Claude Code's; `install` registers them and `/multi-agent:doctor` (`hook-reachability`) says which are missing from your settings.
+
+| Hook | Claude Code | Codex CLI | Copilot CLI |
+| --- | --- | --- | --- |
+| `pre-commit-check.sh` (secret scan, owned paths, attended build/test evidence, unattended ledger) | PreToolUse `Bash` | none: pre-push check | none: pre-push check |
+| `agent-guard.sh` (attribution, force-push, skipped git hooks, argv secrets, closure confirmations, read-only repos, unattended policy) | PreToolUse `Bash`, file edits, fetch and toolkit tools | none | none |
+| `run-scope-guard.sh` (Figma phase rule, worktree boundary warning) | PreToolUse Figma tools, file edits | none: the phase records its own MCP calls | none |
+| `run-stop-guard.sh` (no silent stop mid-run) | Stop | none: halts stay the phase's own write | none |
+| `edit-syntax-gate.sh` | PostToolUse file edits | PostToolUse `apply_patch` | none: Copilot discards postToolUse output |
+| `check-read-size.sh`, capture hooks | registered | none | none |
+
+Where a host has no equivalent hook the rule is still in the phase documents; it is a request there, not enforcement.
+
 ## Outside a pipeline run
 
 Installing the pipeline is not only useful when you run it. Open an ordinary session and the same three things are available, announced by `rules/outside-the-pipeline.md` which loads with every conversation:

package/README.tr.md

@@ -157,7 +157,7 @@ Koşunun kendisinde ayarlanabilen tek şey autopilot; geri kalan her şey kendi
 
 ## Komutlar
 
-`/multi-agent` ve 66 alt komut. `/multi-agent:help` aynı katalogu terminalde, `outputLanguage` ayarına göre gösterir.
+`/multi-agent` ve 69 alt komut. `/multi-agent:help` aynı katalogu terminalde, `outputLanguage` ayarına göre gösterir.
 
 ### Pipeline girişleri
 
@@ -407,17 +407,17 @@ Bu, ilgili plugin'i (+ ortak `ai-common` plugin'ini) repo'nun `.claude/settings.
 
 ## Araç desteği
 
-Pipeline **Claude Code**, **Copilot CLI** ve **Codex CLI** üzerinde native çalışır - üçü de aynı `pipeline/` kaynağından kurulur ve aynı 66 komutu alır.
+Pipeline **Claude Code**, **Copilot CLI** ve **Codex CLI** üzerinde native çalışır - üçü de aynı `pipeline/` kaynağından kurulur ve aynı 69 komutu alır.
 
 | Araç        | Bayrak                  | Ne kurar                                                                                                         |
 | ----------- | ----------------------- | ---------------------------------------------------------------------------------------------------------------- |
 | Claude Code | `--claude` (varsayılan) | slash komutları + skill'ler + agent'lar + üç `PreToolUse` hook'u (secret scan, agent-guard, okuma-boyutu geçidi) |
-| Copilot CLI | `--copilot`             | talimatlar + 66 alt-komut skill'i + script'ler                                                                   |
-| Codex CLI   | `--codex`               | bir router skill + ref olarak 66 spec + 11 agent TOML + `AGENTS.md` bloğu + `codex mcp add`                       |
+| Copilot CLI | `--copilot`             | talimatlar + 69 alt-komut skill'i + script'ler                                                                   |
+| Codex CLI   | `--codex`               | bir router skill + ref olarak 69 spec + 11 agent TOML + `AGENTS.md` bloğu + `codex mcp add`                       |
 
 Skill'leri stack'e göre filtrele: `--platform=ios\|android\|all`.
 
-**Codex neden 66 değil de tek bir skill alıyor.** Codex, keşfettiği her skill'in adını
+**Codex neden 69 değil de tek bir skill alıyor.** Codex, keşfettiği her skill'in adını
 ve açıklamasını tek bir prompt bloğuna toplar ve blok taştığında girdileri hatasızca
 düşürür. 0.145 üzerinde ölçüldü: 142 skill deklare eden bir plugin kurulduğunda sadece
 75'i yüzeye çıktı ve alakasız bir kullanıcı skill'i tahliye edildi. Bu yüzden Codex'te

package/docs/architecture.md

@@ -113,7 +113,7 @@ graph TB
     end
 
     subgraph "Pipeline Specs"
-        CMD[commands/<br/>66 command files]
+        CMD[commands/<br/>69 command files]
         AGT[agents/<br/>11 agent personas]
         RUL[rules/<br/>13 domain rules]
         PHS[multi-agent-refs/phases/<br/>phase specs + contracts]
@@ -204,8 +204,8 @@ repo and the website, plus the two independently shipped repos
 ```mermaid
 graph TD
     CC["Claude Code<br/>(source of truth)"]
-    COP["Copilot CLI<br/>(instructions + 66 sub-command skills)"]
-    COD["Codex CLI<br/>(1 router skill + 66 refs)"]
+    COP["Copilot CLI<br/>(instructions + 69 sub-command skills)"]
+    COD["Codex CLI<br/>(1 router skill + 69 refs)"]
     REPO["Pipeline Repo<br/>(npm package)"]
     WEB["Website"]
     PLUGREPO["multi-agent-plugins<br/>(6 plugins, own repo)"]

package/docs/ecosystem.md

@@ -5,7 +5,7 @@ separately, wired together at install time and at run time:
 
 | Repo | What it owns | Ships as |
 |---|---|---|
-| **`multi-agent-pipeline`** (this repo) | Orchestration: the 6-phase flow, the 66 slash commands, quality gates, review/triage, cross-CLI parity | npm package (`@mmerterden/multi-agent-pipeline`), installs itself onto Claude Code / Copilot CLI / Codex CLI |
+| **`multi-agent-pipeline`** (this repo) | Orchestration: the 6-phase flow, the 69 slash commands, quality gates, review/triage, cross-CLI parity | npm package (`@mmerterden/multi-agent-pipeline`), installs itself onto Claude Code / Copilot CLI / Codex CLI |
 | **`multi-agent-plugins`** | Stack knowledge: per-platform component/lifecycle skills (iOS, Android, Frontend, Backend) + shared knowledge | Claude Code marketplace, 6 independently-versioned plugins |
 | **`multi-agent-toolkit-mcp`** | The pipeline's hands on devices and browsers: 197 MCP tools across 21 categories (simulator/emulator control, end-to-end locators and assertions, memory, crash diagnostics, accessibility audit, store compliance, web automation and scraping, Figma-vs-mock design audit, code intelligence, wallet passes, an agent-DSL batch runner, a full-text context index with text-to-image rendering, provider-backed research, verified hand-off of simple text tasks to free models, video key frames, offline security scoring, read-only access to related repositories at their latest branch, travel document fixtures, a voice channel, App Store Connect and Play Console APIs, virtual iPhones) | npm package, registered as a standard stdio MCP server on every host |
 
@@ -18,7 +18,7 @@ Either can be swapped or removed without touching the other two's source.
 graph LR
     subgraph PIPE ["multi-agent-pipeline (orchestrator)"]
         direction TB
-        PHASES["6 phases · 66 commands"]
+        PHASES["6 phases · 69 commands"]
         GATES["deterministic gates + review triage"]
     end
 
@@ -68,8 +68,8 @@ only those:
 graph TD
     CC["Claude Code<br/>~/.claude/commands/multi-agent/<br/>(source of truth)"]
 
-    CC -->|"Step 2: copy + reformat<br/>66 sub-command skills"| COP["Copilot CLI<br/>~/.copilot/skills/"]
-    CC -->|"Step 2b: transform<br/>(install.js --codex)"| COD["Codex CLI<br/>1 router skill + 66 refs<br/>+ 11 agent TOML"]
+    CC -->|"Step 2: copy + reformat<br/>69 sub-command skills"| COP["Copilot CLI<br/>~/.copilot/skills/"]
+    CC -->|"Step 2b: transform<br/>(install.js --codex)"| COD["Codex CLI<br/>1 router skill + 69 refs<br/>+ 11 agent TOML"]
     CC -->|"Step 3: genericize<br/>(strip personal data)"| REPO["multi-agent-pipeline repo<br/>pipeline/"]
     CC -->|"Step 4: version + feature sync"| WEB["Website<br/>projects.ts / i18n.tsx"]
 
@@ -161,7 +161,7 @@ measurements behind this table):
 
 | | Claude Code | Copilot CLI | Codex CLI |
 |---|---|---|---|
-| **Pipeline commands** | 66 slash-command skills, native | 66 sub-command skills, `multi-agent-{cmd}` naming, copied in | 1 router skill (`multi-agent`) + 66 command specs as reference files - Codex silently truncates its skills block past a few dozen entries, so sub-commands are not peer skills here |
+| **Pipeline commands** | 69 slash-command skills, native | 69 sub-command skills, `multi-agent-{cmd}` naming, copied in | 1 router skill (`multi-agent`) + 69 command specs as reference files - Codex silently truncates its skills block past a few dozen entries, so sub-commands are not peer skills here |
 | **Stack plugins** | Marketplace plugin, loaded natively, resolved by `.claude/settings.json` enabled-list | Enabled plugin's authored skills copied flat into `~/.copilot/skills/`; `knowledge/` **not** re-copied (already delivered via `shared/external`) | Copied as reference files under `~/.codex/multi-agent-refs/skills/`, plugin-prefixed on name clash (e.g. `architecture` → `ai-ios-toolkit-architecture`) |
 | **Component dispatch (Phase 3)** | Marketplace plugin's `create-component`/`create-screen` skill via the Skill tool | No plugin loader - the enabled stack plugin's authored skills (incl. `create-component`) are copied flat into `~/.copilot/skills/` at install time (the old frozen `figma-*` copies are pruned, they were never a fallback) | Not part of the enforced parity axis; classification + state-shape must match, skill *inventory* does not |
 | **multi-agent-toolkit-mcp** | `claude mcp add multi-agent-toolkit -- npx -y @mmerterden/multi-agent-toolkit-mcp` | `copilot mcp add multi-agent-toolkit -- npx -y @mmerterden/multi-agent-toolkit-mcp` | `codex mcp add multi-agent-toolkit -- npx -y @mmerterden/multi-agent-toolkit-mcp` (skipped with a warning if `codex` isn't on `PATH`) |

package/docs/facts.json

@@ -1,7 +1,7 @@
 {
   "$comment": "Generated by pipeline/scripts/gen-facts.mjs. Do not hand-edit: the site reads this, and a number edited here instead of at its source is the drift this file removes.",
   "generatedAt": "2026-10-24",
-  "version": "20.28.0",
+  "version": "20.30.0",
   "phaseSchema": 2,
   "phases": [
     {
@@ -35,22 +35,23 @@
     "autopilot": [0, 1, 2, 3, 4, 5],
     "analysis": [0, 1, 3, 4, 5]
   },
-  "commandCount": 66,
+  "commandCount": 69,
   "skillCount": 165,
-  "skillCountAll": 234,
+  "skillCountAll": 237,
   "agentCount": 11,
   "hosts": ["Claude Code", "Codex CLI", "Copilot CLI"],
-  "toolCount": 204,
+  "toolCount": 208,
   "toolCategories": {
     "ios": 42,
     "web": 38,
-    "android": 33,
+    "android": 34,
     "store": 30,
     "vphone": 10,
     "code": 8,
     "design": 7,
     "repo": 6,
     "voice": 6,
+    "ui": 5,
     "context": 4,
     "pass": 4,
     "document": 3,
@@ -58,11 +59,10 @@
     "llm": 2,
     "research": 2,
     "security": 2,
-    "ui": 2,
     "coverage": 1,
     "media": 1,
     "test": 1
   },
   "pluginSkillCount": 261,
-  "toolkitVersion": "3.25.0"
+  "toolkitVersion": "3.26.0"
 }