chore: sync private v4.8.9 (99ad2c3)

.cursor-plugin/plugin.json

@@ -13,6 +13,6 @@
     "anti-slop"
   ],
   "license": "MIT",
-  "version": "4.8.4",
+  "version": "4.8.9",
   "repository": "https://github.com/agent-kit-startup/agent-kit"
 }

.cursor/agent-kit.json

@@ -1,6 +1,6 @@
 {
   "schemaVersion": 1,
-  "version": "4.8.4",
+  "version": "4.8.9",
   "protected": [
     ".cursor/HANDOFF.md",
     ".cursor/agents/test-suites.md",

.cursor/commands/agent-kit-onboard.md

@@ -100,6 +100,6 @@ When complete:
    - `Next: /start-project` when the user wants to plan a deliverable.
    - `Next: finish setup` when no deliverable should start now.
 
-After essentials are ready, Mission Control is **optional** and **not** an essential readiness check. Consumer L0 installs the `/dashboard` command text but not `dashboard/**`. If the operator wants the panel, point them to an agent-kit checkout that includes `dashboard/start.mjs` (loopback `http://127.0.0.1:3333`). Do not block `/start-project` on Mission Control. Do not ask about skins or external review before essentials (Hard Stop 1).
+After essentials are ready, Mission Control is **optional** and **not** an essential readiness check. Consumer L0 installs the `/dashboard` command text but not `dashboard/**`. If the operator wants the panel, `agent-kit dashboard` serves it from the installed CLI (4.8.2 onward); on older pins point them to an agent-kit checkout that includes `dashboard/start.mjs` (loopback `http://127.0.0.1:3333`). Do not block `/start-project` on Mission Control. Do not ask about skins or external review before essentials (Hard Stop 1).
 
 Agent Personas remain available through later personalization or settings. External review is offered only when a plan reaches exhaustion.

.cursor/commands/backlog-add.md

@@ -120,3 +120,4 @@ Vague-goal clarify and write confirmation **must** use Ask questions per `.curso
 
 - ADR: `.cursor/memory/decisions/2026-07-26_backlog-crud-commands-contract.md`
 - Disposition gate for `/start-project`: `.cursor/memory/decisions/2026-07-25_start-project-plan-disposition-gate.md`
+- Cursor product-update gaps may route here via `/cursor-update-awareness` (Ask → `/backlog-add`)

.cursor/commands/cursor-update-awareness.md

@@ -0,0 +1,64 @@
+# Command: /cursor-update-awareness
+
+## Goal
+
+Run an **opt-in advisory** check for Cursor product updates (changelog + `docs/cursor-native-audit.md` inventory), then route **confirmed** gaps into the existing conveyor with HITL. Never apply kit/IDE changes. Never auto-create Field Reports or public issues.
+
+**Detection source ADR:** `.cursor/memory/decisions/2026-08-01_cursor-update-detection-source.md` (changelog fetch primary; sessionStart = delivery; readiness = optional storage only).
+
+## When to Use
+
+- After a Cursor IDE upgrade or when `cursorUpdateCheck.enabled` sessionStart nudge mentions gaps
+- When reviewing whether Agent Kit should adopt a new Cursor surface (hooks, MCP, skills, commands, SDK)
+- Before `/backlog-add` or `/dogfood` for Cursor-integration work
+
+## Hard stops
+
+1. **Check ≠ apply.** CLI and this command only report. No silent rewrite of `.cursor/`, no Marketplace submit, no native-audit version-prose refresh (owned by parked `submit-cursor-marketplace`).
+2. **No auto Field Reports** and no auto GitHub issues.
+3. **Lane separation:** `/dogfood` stays factory vs consumer aware (ADR `2026-07-31_dogfood-factory-consumer-lanes.md`). Do not write consumer notes into factory `dogfood/` without an explicit operator bridge.
+4. **HITL before enqueue.** Confirmed gaps go through Ask → `/backlog-add` or Ask → `/dogfood`, never silent backlog rows.
+5. **Never `/git-prod`.**
+
+## Check-only (CLI)
+
+```bash
+agent-kit cursor-awareness --check [--json] [--respect-prefs] [--stamp] [--offline]
+```
+
+- Fetches `https://cursor.com/changelog` (override via `cursorUpdateCheck.changelogUrl`) unless `--offline`.
+- Diffs against `docs/cursor-native-audit.md` (open Action items, refresh staleness) and validates `docs/cursor-3-features.md` presence.
+- Prefs in `.cursor/context/config.json` under `cursorUpdateCheck` (`enabled` default `false`, `intervalDays`, `lastSeenCursorVersion`). Distinct from kit `updateCheck`.
+- `applyRecommended` and `fieldReportRecommended` are always `false`.
+
+## What to Do
+
+1. **Run the check** (prefer CLI JSON):
+   ```bash
+   agent-kit cursor-awareness --check --json
+   ```
+   Fallback: inventory-only `agent-kit cursor-awareness --check --offline --json`.
+
+2. **Summarize gaps** for the operator (id, severity, evidence, suggestedRoute). If status is `current` or `skipped-*`, report and stop.
+
+3. **Confirm routing via Ask questions** (chat numbered-list fallback). One question:
+
+   > Cursor awareness found N advisory gap(s). Route confirmed work?
+
+   Options (labels exact):
+   - `Enqueue via /backlog-add`
+   - `File /dogfood note`
+   - `Not now`
+
+4. **Handlers:**
+   - `Enqueue via /backlog-add`: hand off to `/backlog-add` with a goal summarizing the confirmed gaps (Broad Intake + write Ask still apply). Do not activate or run the new plan.
+   - `File /dogfood note`: hand off to `/dogfood` with a hygiene-stripped topic/summary. Respect factory vs consumer lane.
+   - `Not now` / skipped: stop. Optionally offer enabling `cursorUpdateCheck.enabled` for future sessionStart nudges (do not mutate config without Ask).
+
+5. **Out of scope:** refreshing Marketplace plugin version prose; kit self-release (`/update` / `updateCheck`); auto-remediation of product code.
+
+## Related
+
+- Kit consumer autoupdate (separate): `/update`, ADR `2026-07-27_consumer-autoupdate-check-opt-in.md`
+- Dogfood ingest: `/dogfood`, ADR `2026-07-31_dogfood-ingest-contract.md`
+- Triage residuals path: `/plan-review-triage` → Write residuals → `/backlog-add`

.cursor/commands/dashboard.md

@@ -6,7 +6,7 @@ Start (or reuse) Mission Control for **this Cursor workspace only**, then open t
 
 Local-dev only. Read-only. No HITL gate.
 
-**Terminal counterpart:** `agent-kit dashboard` from the workspace cwd. After a CLI publish that ships Path C, the installed package includes `dashboard/start.mjs`. Fallbacks: env `MISSION_CONTROL_KIT_ROOT` / `AGENT_KIT_HOME`, sibling `../agent-kit`, or `node "$KIT_ROOT/dashboard/start.mjs"` with `MISSION_CONTROL_REPO_ROOT` set to this git root. Until Path C is on npm, do not assume `@dadado/agent-kit-cli@4.8.0` has the panel assets.
+**Terminal counterpart:** `agent-kit dashboard` from the workspace cwd. The installed package includes `dashboard/start.mjs` from 4.8.2 onward. Fallbacks: env `MISSION_CONTROL_KIT_ROOT` / `AGENT_KIT_HOME`, sibling `../agent-kit`, or `node "$KIT_ROOT/dashboard/start.mjs"` with `MISSION_CONTROL_REPO_ROOT` set to this git root. On 4.8.0 or an older pin the panel assets are absent.
 
 ## When to Use
 

.cursor/commands/dogfood.md

@@ -0,0 +1,138 @@
+# Command: /dogfood
+
+## Goal
+
+File a private dogfood note from the current chat or explicit arguments into the local inbox. Factory (`agent-kit-dev`) writes to `dogfood/`; consumer projects write to `.cursor/dogfood/`. Never syncs upstream; never creates a public issue automatically.
+
+## When to Use
+
+- You hit Agent Kit friction, a bug, or a surprising behavior and want it tracked for internal analysis.
+- You want a dated record of a session pattern without turning it into a public issue or PR.
+- A memory WRITE is not yet warranted; the note is raw material for later triage.
+
+## Usage
+
+```
+/dogfood <topic> [one-line summary]
+```
+
+Examples:
+- `/dogfood plan-handoff context lost after long run`
+- `/dogfood update-refresh personalization dropped`
+- `/dogfood stale-pack-id install failed on pack migration`
+
+Without arguments, summarize the current chat turn into the topic and body.
+
+## Hard stops
+
+1. **Detect lane before writing.**
+   - **Factory lane** — this checkout is the canonical `agent-kit-dev` repository (`origin` remote contains `agent-kit-dev`, or `dogfood/` already exists at repo root). Write to `dogfood/cursor_<topic>_<date>.md` and update `dogfood/README.md` Unprocessed index.
+   - **Consumer lane** — any other project with an Agent Kit install (`.cursor/agent-kit.json` exists). Write to `.cursor/dogfood/cursor_<topic>_<date>.md` and a local index. Do **not** track the folder in git; it should already be gitignored by the base install.
+   - **Unknown lane** — stop and ask the operator which lane to use.
+2. **Hygiene strip (mandatory).** Remove before writing:
+   - Consumer workspace names, project names, domain names, or external product names.
+   - People's names, Slack channels, client IDs, or organization names.
+   - Chat-transient phrasing ("as I mentioned", "dear user", "conforme falamos").
+   - Any value that looks like a secret, token, or credential.
+3. **Session origin ≠ product use case.** The note describes an Agent Kit system pattern, not a specific consumer's business process. See `2026-07-17_session-origin-not-product-usecase.md`.
+4. **No public issue or PR without explicit HITL.** If the operator wants a public issue, use `/dogfood` → local save first, then offer a separate `/contribute` or `gh issue create` step after the hygiene strip is verified.
+5. **No Field Report cards.** Routine dogfood filing does not create or update Field Report cadence.
+
+## What to Do
+
+### Step 1: Determine lane
+
+Check in order:
+1. `git remote get-url origin` (or `git config remote.origin.url`) contains `agent-kit-dev` → factory lane.
+2. A `dogfood/` directory exists at the repo root and contains this README → factory lane.
+3. `.cursor/agent-kit.json` exists and the repo is not `agent-kit-dev` → consumer lane.
+4. Neither → stop and ask: "This doesn't look like the factory or a migrated consumer. Save to `dogfood/` (factory) or `.cursor/dogfood/` (consumer)?"
+
+### Step 2: Build filename and body
+
+1. Normalize the topic: lowercase, spaces/hyphens/underscores to underscores, strip punctuation and trailing date. Keep it short (≤ 40 chars).
+2. Date suffix: `YYYYMMDD` from today (`YYYY_MM_DD` for readability).
+3. Filename: `cursor_<topic>_<date>.md`.
+4. Body template:
+   ```markdown
+   # Dogfood: <topic>
+
+   - **Date:** <YYYY-MM-DD>
+   - **Lane:** factory | consumer
+   - **Source:** chat summary | explicit /dogfood args
+
+   ## Observation
+
+   <hygiene-stripped description of the friction>
+
+   ## Impact
+
+   <how this affects the Agent Kit system or operator experience>
+
+   ## Triage (initial)
+
+   - Fix now / Park / Ignore
+   - **Tags:** <lowercase comma-separated keywords>
+   ```
+5. Run the hygiene strip. If you cannot strip enough context to make it generic, file the note but flag it as `needs-anonymization` in the body and stop before any memory WRITE or public issue.
+
+### Step 3: Write and index
+
+**Factory lane:**
+- Write `dogfood/cursor_<topic>_<date>.md`.
+- Append the file to the `### Unprocessed Files` section of `dogfood/README.md` with a one-line summary and the capture date.
+
+**Consumer lane:**
+- Ensure `.cursor/dogfood/` exists (create if missing).
+- Write `.cursor/dogfood/cursor_<topic>_<date>.md`.
+- Write or append to `.cursor/dogfood/README.md` with the same Unprocessed/Processed structure as the factory README.
+- The folder is local-only; do not `git add` it.
+
+### Step 4: Cross-repo bridge (optional, operator-initiated only)
+
+A consumer project does **not** write directly into the factory repo. If the operator wants a note from `.cursor/dogfood/` to reach the canonical `agent-kit-dev` inbox, the supported path is a manual bridge:
+
+1. **Configure factory root** (optional). In `.cursor/context/config.json` add:
+   ```json
+   {
+     "dogfood": {
+       "factoryRoot": "/absolute/path/to/agent-kit-dev"
+     }
+   }
+   ```
+   The path is advisory only; the command never writes there automatically.
+2. **Operator copies the file** with `cp` or the IDE file explorer from `.cursor/dogfood/cursor_<topic>_<date>.md` to `dogfood/cursor_<topic>_<date>.md` in the factory checkout.
+3. **Re-apply hygiene** in the factory context before committing. The file must be re-reviewed because the factory README index and triage cycle are separate from the consumer inbox.
+4. **No bridge for routine friction.** Most consumer notes should stay in `.cursor/dogfood/` as local project memory. Only copy patterns that are clearly Agent Kit system gaps.
+
+If `dogfood.factoryRoot` is absent, omit the bridge step and file locally. Never invent a factory path or guess from repo history.
+
+### Step 5: Optional public issue (HITL only)
+
+After filing locally, you may offer a public GitHub issue **only if**:
+1. The pattern is an upstream Agent Kit gap, not a project-specific workaround.
+2. The hygiene strip has already been applied (no consumer identities, no session-origin detail).
+3. The operator explicitly agrees via Ask questions with options:
+   - `Create public issue`
+   - `Keep local only`
+
+If the operator chooses `Create public issue`:
+- Use `gh issue create` against the public `agent-kit-startup/agent-kit` repository.
+- Title format: `[Dogfood] <topic>`.
+- Body: a concise, anonymized summary plus the local dogfood file path for reference.
+- Do **not** paste the full local dogfood file if it contains any non-public detail.
+- Never create a Field Report card or cadence warning for this step.
+
+If the operator chooses `Keep local only`, stop. The local file is the record.
+
+### Step 6: Respond
+
+> Dogfood filed: `dogfood/cursor_<topic>_<date>.md` (factory) or `.cursor/dogfood/cursor_<topic>_<date>.md` (consumer). Next: analyze → memory WRITE → triage, or ask for a public issue if the pattern is upstream-relevant.
+
+## Related
+
+- `dogfood/README.md` — factory inbox and ingest ritual
+- `.cursor/memory/decisions/2026-07-31_dogfood-factory-consumer-lanes.md` — lane decision
+- `.cursor/memory/decisions/2026-07-31_dogfood-ingest-contract.md` — ingest contract
+- `.cursor/memory/decisions/2026-07-17_session-origin-not-product-usecase.md` — hygiene
+- Cursor product-update gaps may route here via `/cursor-update-awareness` (Ask → `/dogfood`)

.cursor/commands/git-staging.md

@@ -6,7 +6,7 @@ Follow the **git staging** routine to bring local changes to the pre-production
 
 1. **Read** the "Prompt: git staging" section in `autogit/gitupdate.md` (when it exists).
 2. **Staging hygiene (monitors):** if `git status` shows untracked or unrelated dirty `.cursor/memory/plan-monitor-*.md`, **warn** before commit. Stage memory/monitor files **add-by-name only**; never broad `git add` of `.cursor/memory/` WIP into a product commit (ADR `decisions/2026-07-27_plan-monitor-consumer-awareness.md`, external-review staging hygiene).
-3. **Lint evidence (required when code/format paths change):** before claiming staging-ready, **run** the repo formatter/linter on touched files and **record the command + result** (pass/fail) in the worker summary or tick notes. Writing `Staging ready: yes` or the contract string alone is **not** evidence. Pure markdown / docs-only with no applicable linter: state `none applicable`. Same gate as `/run-plan` Staging-ready lint gate.
+3. **Lint evidence (required when code/format paths change):** before claiming staging-ready, **run** the repo formatter/linter on touched files and **record the command + result** (pass/fail) in the worker summary or tick notes. Writing `Staging ready: yes` or the contract string alone is **not** evidence. Pure markdown / docs-only with no applicable linter: state `none applicable`. Same gate as `/run-plan` Staging-ready lint gate. **Dashboard CSS/HTML only** (`dashboard/dashboard.html` and similar, outside Biome scope): record `Tests: none applicable (dashboard-CSS); covered by plugin-ux-validation` when the UX suite pins the change (ADR `decisions/2026-07-29_dashboard-css-lint-evidence-convention.md`); do not claim Biome covered the HTML.
 4. Run in order: validation (not on `main`), CHANGELOG (`[Unreleased]`), checkout staging, pull, working branch, Conventional Commits, push, MR/PR (**always `--base staging` / target `staging`**), merge, cleanup.
 5. **Never** commit directly to `main`.
 6. On completion: update `.cursor/HANDOFF.md` (phase in staging); memory-loop WRITE if it applies.

.cursor/commands/plan-external-review.md

@@ -47,7 +47,8 @@ If any are missing: stop. Do **not** claim a review ran. Tell the user to run `a
 ### What "operator-visible" means (smoke notes)
 
 - **Autonomous success:** chat arm **must** use `--force --autonomous --wait-monitor`. The launcher prefers a background/inspectable PTY (no OS Terminal focus by default; `--focus-terminal` / `AGENT_KIT_AUDIT_FOCUS_TERMINAL=1` restores activate), then polls until a **fresh** monitor exists (`mtime >= arm epoch` or content sentinel). Exit `0` = fresh ready; `3` = timeout; `4` = soft-fail while waiting. Spawn-only exit 0 without wait is **not** review done. **Chat continuation:** AwaitShell until `0|3|4`; on `0` run `/plan-review-triage` Ask in the same session. Do **not** stop at Final HANDOFF "after monitor lands" or require typing `done`. ADR: `decisions/2026-07-27_audits-wait-freshness-enforce.md`.
-- **Autonomous soft-fail:** missing `claude` → tip + exit `4` when `--wait-monitor` was requested (Field Report owed). Background spawn unavailable → falls back to `--paste-only` UX with an honest "NOT running yet" banner. Soft-fail does **not** invent a monitor or run triage as if review completed.
+- **Autonomous soft-fail:** missing `claude` → tip + exit `4` when `--wait-monitor` was requested (Field Report owed). Background spawn unavailable → falls back to `--paste-only` UX with an honest "NOT running yet" banner. A **silent PTY** (spawn succeeded, no scrollback within the progress-gate grace window) is reported as a failed launch: the launcher disposes the session it just spawned, prints the paste fallback, and soft-fails instead of burning the wait budget. A **session-cap refusal** (detached `agent-kit-audit-*` sessions at the cap) never spawns at all. Soft-fail does **not** invent a monitor or run triage as if review completed.
+- **Exit 3 is timeout-only:** it means the freshness gate was not satisfied inside the budget, never that the review finished. A monitor that appears later, including one written by a different or later arm, does **not** convert a `3` into success. Leave the target Field Report **owed** and re-arm. ADR: `decisions/2026-07-30_audits-pty-progress-gate-zombie-policy.md`.
 - **Paste-only:** clipboard + printed interactive one-liner; review starts only after the operator pastes into their Cursor Terminal. After paste (Claude running), the session still waits for the monitor file then continues into triage Ask when possible.
 - **`--dry-run`:** resolves mode/plan and prints `background-cmd` / `paste-cmd` / `focus-terminal` without spawning Claude (useful for smoke).
 
@@ -92,6 +93,8 @@ Script behavior (ADR):
 - `mode: "autonomous"` (non-headless) → background/inspectable PTY auto-launch; soft-fallback to paste-only
 - Missing `mode` key → paste-compatible default (`--print` when no flag; chat should pass `--paste-only` or set autonomous)
 - Interactive and headless launches pass Claude CLI `--permission-mode auto`
+- Post-spawn progress gate: samples PTY scrollback before the monitor wait; silent PTY → early abort (`AGENT_KIT_AUDIT_PROGRESS_TIMEOUT`, default 60s, `0` disables); channels without a scrollback API stay advisory
+- Session pressure: warns at `AGENT_KIT_AUDIT_SESSION_WARN` detached `agent-kit-audit-*` sessions, refuses to spawn at `AGENT_KIT_AUDIT_SESSION_CAP`; reap is opt-in (`--reap-audit-sessions`), detached-only, past `AGENT_KIT_AUDIT_REAP_MIN_AGE`
 - `--paste-only` copies the interactive one-liner via `pbcopy` / `xclip` / `xsel` / `clip.exe` when available
 - Never `/git-prod`; never broad `git add`
 - Does **not** register a Cursor native `stop` hook
@@ -120,6 +123,7 @@ After a successful autonomous arm in chat, **do not** hand off with "run `/plan-
 - ADR: `.cursor/memory/decisions/2026-07-20_optional-claude-code-plan-review.md`
 - Audits contract: `.cursor/memory/decisions/2026-07-27_audits-autonomous-plan-review-contract.md`
 - Post-spawn watch + continue: `.cursor/memory/decisions/2026-07-27_audits-post-spawn-monitor-watch-continue.md`
+- PTY progress gate, session cap, exit 3 honesty: `.cursor/memory/decisions/2026-07-30_audits-pty-progress-gate-zombie-policy.md`
 - Related: `.cursor/memory/decisions/2026-07-19_stop-hook-no-hitl-interference.md` (no stop-hook auto agent)
 - Prompt: `.cursor/context/templates/plan-external-review-prompt.md`
 - Monitor template: `.cursor/context/templates/plan-monitor.md`