status

name	status
user-invocable	true
description	Morning briefing — architecture state, feature pipeline, production health, user journeys, git pulse, and daily focus recommendations. Not an audit; a lay of the land.
argument-hint	[--quick|--full] [--focus pipeline|health|journeys|git|agent]
version	1.1.0
status	active
packages	["all"]
dependencies	[]
last_updated	2026-04-07
last_verified	2026-04-07

Status Skill

Situational awareness briefing for starting a work session. Aggregates architecture state, feature pipeline progress, production health signals, user journey readiness, git activity, and recommends what to focus on.

This is NOT an audit. It does not look for problems to fix. It maps the terrain so you can navigate it.

---

Invocation

/status                    # Standard briefing (all sections)
/status --quick            # Abbreviated — pipeline + health + focus only
/status --full             # Deep briefing with subagent exploration
/status --focus pipeline   # Feature pipeline only
/status --focus health     # Production health only
/status --focus journeys   # User journey readiness only
/status --focus git        # Git activity pulse only
/status --focus agent      # Agent & skill ecosystem only

---

Execution Model

• Read-only: Never edit files. All output goes to chat.
• Speed over depth: Favor fast reads over exhaustive scans. This runs at session start — keep it under 2 minutes for standard, 30 seconds for --quick.
• Parallel gathering: Sections 1-6 can be gathered in parallel. The Headline, Blockers, Delta, and Daily Focus synthesize the others.
• For --full mode: Use Explore subagents for deeper architecture and journey mapping. Standard mode uses direct tool calls only.
• Wait gate for --full mode: When subagents are spawned, they MUST run in foreground (NOT run_in_background). Do NOT render the briefing until all subagent results have returned. Incorrect early output with later corrections is worse than a slower correct output.

---

Output Structure

The briefing always follows this order:

1. Headline — one-sentence state of the product (synthesized after gathering)
2. Blockers — anything blocking work, pulled to the top (if any exist)
3. Last Session — continuity from session-state.md (if it exists)
4. Delta — what changed since last /status run (if prior data exists)
5. Sections 1-6 (data)
6. Section 7: Daily Focus (synthesis)

---

Headline

A single sentence at the very top summarizing the product state. Written after all sections are gathered. Should be scannable in under 3 seconds.

Format

> **[N/M journeys operational]. [N features in progress, N dormant]. [Health signal]. [Agent pipeline signal].**

This is a template — fill from gathered data. Never reuse example values verbatim.

Rules

• One sentence, max two. No bullet points.
• Cover: which journeys work, what's mock/broken, pipeline momentum, health signal.
• Use plain language — this is for a human glancing at their terminal first thing in the morning.

---

Blockers Callout

If ANY of the following are true, surface them in a prominent callout block before the sections. Nothing buried = nothing missed.

Sources

• Lanes with status blocked in any feature pack
• Failing health checks (types, lint, tests)
• Broken user journeys (✗ status)
• Stale dependencies blocking a feature
• Uncommitted work on a branch that's diverged significantly

Output

> **⚠ Blockers**
> - **autoresearch/ui** lane blocked — waiting on runtime merge
> - **Tests FAIL** — 2 regressions since last run
> - **Capture (media)** journey broken — receiver handler missing

If nothing is blocked, omit this section entirely. Don't say "No blockers" — absence means clear.

---

Last Session

If session-state.md exists in the repo root, read it and surface a continuity summary.

Gather

1. Check if session-state.md exists at repo root
2. If yes, read it and extract: current task, progress, files modified, next steps, blockers

Output

> **Continuing from last session** (saved 2026-04-06T23:15:00Z)
> Task: Autoresearch QA pass 2
> Left off: experiment-loop tests passing, variant-generation untested
> Next: Wire variant UI into Nest tab settings

If no session-state.md exists, omit this section. Don't mention its absence.

---

Delta (Change Since Last Status)

Track what moved between status runs. This turns the briefing from a snapshot into a momentum tracker.

Gather

After all sections are gathered, compare current values against the previous run's data. The previous run's data comes from:

1. The most recent /status output in conversation history (same session)
2. Or absence — if this is the first run, skip this section

Output

> **Since last status** (2h ago)
> - Tests: 47 → 52 (+5)
> - Skills with eval: 3 → 5 (+2)
> - Branches: -2 merged, +1 new
> - Features: agent-knowledge-sandbox moved planned → in-progress
> - Health: all checks still passing

Rules

• Only show lines where something changed. Don't repeat stable values.
• If this is the first /status in the session, omit this section entirely.
• Use → arrows for transitions, +N/-N for numeric changes.

---

Freshness Indicators

All dates in the briefing should use relative age labels instead of (or alongside) absolute dates. This makes staleness immediately visible.

Age Labels

Age	Label	Urgency
Today	today	Current
1 day	yesterday	Current
2-6 days	Nd ago	Normal
7-13 days	~Nw ago	Attention
14+ days	Nd ago ⚠ stale	Stale — flag it

Application

Use these everywhere dates appear:

• Feature pipeline Updated: field → Updated: today or Updated: 12d ago ⚠ stale
• Git pulse timestamps
• Session state saved date
• Build staleness

---

Section 1: Architecture Snapshot

Quick structural overview — not a deep analysis (that's /architecture).

Gather

1. Read packages/*/package.json — extract name, version, key dependency versions
2. Count shared modules: ls packages/shared/src/modules/
3. Count extension views: ls packages/extension/src/views/
4. Count skills: ls packages/extension/src/skills/
5. Check @coop/shared barrel: count exports in packages/shared/src/index.ts

Output

## Architecture Snapshot

| Package | Version | Key Deps |
|---------|---------|----------|
| @coop/shared | x.x.x | viem, yjs, dexie, zod |
| @coop/app | x.x.x | react, vite |
| @coop/extension | x.x.x | wxt, react, webllm |
| @coop/api | x.x.x | hono, bun |

Shared modules: N  |  Extension views: N  |  Agent skills: N
Shared barrel exports: ~N

---

Section 2: Feature Pipeline

The heart of the briefing. Read all feature packs and present where everything stands.

Gather

1. Read every .plans/features/*/status.json that exists
2. For features without status.json, read the spec.md header to infer stage
3. Check for any .plans/*.todo.md flat files (legacy)
4. Read .plans/features/*/lanes/*.todo.md frontmatter for lane-level status

Classification

Group features by lifecycle stage:

Stage	Criteria
Shipped	All lanes done, merged to main
In Progress	At least one lane in_progress or done with others remaining
Ready	All lanes todo or ready, no blockers
Planned	Has spec but lanes not started
Blocked	Has explicit blockers or unresolved dependencies
Dormant	No status.json, updated_at > 14 days old

Output

## Feature Pipeline

### In Progress
- **autoresearch** — Self-optimizing agent skills
  Lanes: eval ✓ | state ✓ | runtime ✓ | ui ✓ | qa1 ✓ | qa2 ✓
  Updated: today

### Planned
- **agent-knowledge-sandbox** — Curated sources + graph memory
  Lanes: eval → | state → | runtime → | ui →
  Updated: yesterday

### Dormant
- **media-compression-sharing** — 12d ago ⚠ stale
- **receiver-design-polish** — 12d ago ⚠ stale

Total: N features (N shipped, N in progress, N planned, N dormant)

Use lane indicators: ✓ done, ◐ in progress, → ready/todo, ✗ blocked, — n/a

---

Section 3: Production Health

Quick signals — not a full monitor run (that's /monitor).

Gather

1. Run bun run validate typecheck 2>&1 | tail -5 — pass/fail
2. Run bun lint 2>&1 | tail -5 — pass/fail + warning count
3. Run bun run test 2>&1 | tail -15 — test count, pass/fail summary
4. Check last successful build: git log --oneline --grep='build' -5 or check if dist/ dirs exist
5. Check environment: read .env.local for current mode settings (mock/live)

Output

## Production Health

| Check | Status | Detail |
|-------|--------|--------|
| Types | PASS | 0 errors |
| Lint | PASS | 2 warnings |
| Tests | PASS | 47/47 |
| Build | STALE | Last build: 2 days ago |

Environment: sepolia / mock onchain / mock archive

Statuses: PASS, WARN (non-blocking issues), FAIL (blocking), STALE (not recently verified), UNKNOWN

---

Section 4: User Journey Map

Which product flows are wired and operational vs stubbed or incomplete.

Gather

For each journey, read the sentinel file(s) listed below. A journey is ● Operational if the sentinel contains real logic (function bodies, API calls, crypto, protocol handling). It is ◐ Partial only if the sentinel exists but contains TODOs, empty stubs, or placeholder returns. It is ○ Mock only if the implementation is a mock/fake that returns hardcoded data. It is ✗ Broken if the file is missing or has import/compile errors. Do not infer status from config mode toggles — a real implementation behind a mock-mode flag is still ● Operational.

#	Journey	Sentinel file(s)	Look for
1	Capture (tabs)	extension/src/background/handlers/capture.ts	extractPageSnapshot, savePageExtract
2	Capture (media)	app/src/views/Receiver/CaptureView.tsx	getUserMedia, MediaRecorder, onPickFile
3	Refine (agent)	extension/src/runtime/agent/index.ts, extension/src/runtime/skills/runner.ts	handleRunAgentCycle, skill registry, WebLLM bridge
4	Review	extension/src/views/Sidepanel/tabs/ChickensTab.tsx	Draft cards, approve/reject, useDraftEditor
5	Share (local sync)	shared/src/modules/storage/index.ts, grep for y-webrtc or connectReceiverSyncProviders	Yjs doc creation, provider wiring, peer transport
6	Share (archive)	shared/src/modules/archive/index.ts	uploadArchiveBundle, createStorachaArchiveClient, UCAN delegation
7	Auth (passkey)	shared/src/modules/auth/index.ts	createWebAuthnCredential, restorePasskeyAccount
8	Auth (Safe)	shared/src/modules/onchain/index.ts	createSmartAccountClient, Safe v1.4.1, ERC-7579
9	Sync (signaling)	api/src/ws/ or api/src/index.ts	WebSocket upgrade, room management, awareness protocol

All sentinel paths are relative to packages/. In standard mode, read the first 50 lines of each sentinel. In --full mode, the Explore subagent does deeper analysis.

Output

## User Journeys

| Journey | Status | Notes |
|---------|--------|-------|
| Capture (tabs) | [indicator] | [evidence from sentinel read] |
| Capture (media) | [indicator] | [evidence from sentinel read] |
| ... | ... | ... |

Indicators: ● operational, ◐ partial/WIP, ○ mock/stubbed, ✗ broken/missing

Never copy example statuses from this skill definition. Every status must come from reading the actual sentinel files.

---

Section 5: Git Activity Pulse

Recent activity context to understand momentum and open work.

Gather

1. git log --oneline -20 — recent commits
2. git branch -a --sort=-committerdate | head -15 — active branches
3. git status — uncommitted work
4. git stash list — stashed work
5. git log --since='7 days ago' --format='%an' | sort | uniq -c | sort -rn — contributor activity

Output

## Git Pulse

Last 7 days: N commits across N branches
Active branches: feature/xxx, fix/yyy, ...
Uncommitted changes: N files modified, N untracked

Recent commits:
  bdb2c88 docs(agent): define avenger stack and Coop OS architecture
  d5b5155 chore(claude): add orchestration plan
  ...

---

Section 6: Agent & Skill Ecosystem

The agent pipeline is the product. This section surfaces whether the AI layer is healthy.

Gather

1. Skill inventory: ls packages/extension/src/skills/ — count skill directories, note any with missing system.md or empty eval/ dirs
2. Eval fixture health: find packages/extension/src/skills/*/eval -name '*.json' | wc -l — how many eval fixtures exist
3. Agent runtime: Read packages/extension/src/runtime/agent/index.ts header — check for exported entry points, note any TODO/FIXME markers
4. Autoresearch state: Read .plans/features/autoresearch/status.json — are experiments running, what's the latest variant status
5. Skill schema contracts: Check packages/shared/src/contracts/schema-agent.ts for skill-related schemas — count how many are defined
6. Inference cascade: Check packages/extension/src/runtime/agent/experiment-loop.ts — is the loop wired and exported
7. WebLLM readiness: Grep for WebLLM imports/model references to confirm model availability

Output

## Agent & Skill Ecosystem

| Metric | Value |
|--------|-------|
| Skills registered | N |
| Skills with eval fixtures | N / N |
| Agent schemas defined | N |
| Experiment loop | wired / not wired |
| Autoresearch | active / dormant / not started |

### Skill Detail
| Skill | Has system.md | Has eval | Eval fixtures |
|-------|:------------:|:--------:|:-------------:|
| capital-formation-brief | ● | ● | 2 |
| memory-insight-synthesizer | ● | ● | 2 |
| review-dig | ● | ○ | 0 |
| ... | ... | ... | ... |

### Agent Pipeline
- Inference cascade: [wired/partial/missing]
- Variant generation: [active/dormant]
- Last experiment: [date or "none"]

Indicators: ● present, ○ missing

---

Section 7: Daily Focus

Synthesize sections 1-6 into actionable recommendations for the work session.

Logic

1. Blocked features → Unblock first (highest leverage)
2. In-progress features with ready lanes → Continue momentum
3. Failing health checks → Fix before new work
4. Skills missing eval fixtures → Strengthen the agent pipeline
5. Stale builds → Rebuild to establish baseline
6. Dormant features → Decide: continue, defer, or archive
7. Ready features → Pick one to start if nothing else is pressing

Output

## Daily Focus

### Priority
1. **[Action]** — [reason, context]
2. **[Action]** — [reason, context]
3. **[Action]** — [reason, context]

### Consider
- [Lower-priority suggestion]
- [Lower-priority suggestion]

### Parking Lot
- [Dormant items that need a decision but aren't urgent]

Keep to 3 priority items max. Be specific about what to do, not vague.

---

Mode Variants

--quick

Run only:

• Headline (always)
• Blockers callout (if any)
• Last Session (if session-state.md exists)
• Section 2 (Feature Pipeline) — abbreviated, no lane detail
• Section 3 (Production Health) — typecheck only, skip full test run
• Section 7 (Daily Focus) — based on available data

Target: under 30 seconds.

--full

All sections plus:

• Spawn foreground Explore subagents (never background) to map shared module export surface and check user journey wiring depth
• Do NOT render any briefing output until all subagents have returned their results
• Include dependency version audit (outdated packages)
• Include a "What's Changed Since Last Status" diff if a previous status output is found
• Target: under 3 minutes (slower than standard, but correct on first pass)

--focus [section]

Run only the named section:

• pipeline → Section 2
• health → Section 3
• journeys → Section 4
• git → Section 5
• agent → Section 6

---

Output Format

Always output as chat markdown. The briefing should be scannable — use tables, indicators, and short sentences. No prose paragraphs.

Header

# Coop Status — [date]

> **[Headline — one sentence product state summary]**

Immediately followed by Blockers callout (if any), Last Session (if exists), Delta (if not first run), then sections 1-7.

Footer

---
_Briefing generated [timestamp]. For deeper analysis: `/architecture`, `/audit`, `/monitor`._

---

Anti-Patterns

Don't	Why
List every file or export	This is a briefing, not an inventory
Flag issues to fix	This is not an audit — observe, don't prescribe fixes
Run full test suite in --quick mode	Speed matters at session start
Make architectural judgments	Describe the state, don't evaluate it
Edit any files	Read-only, always
Recommend refactoring	That's /architecture territory
Spend more than 2 minutes	If gathering takes too long, report what you have

---

Execution Checklist

When /status is invoked:

1. Parse arguments for mode (--quick, --full, --focus)
2. Check for session-state.md at repo root — read if present
3. Gather in parallel (sections 1-6 as independent data collection)
   • For standard mode: direct tool calls (Read, Bash, Glob, Grep). For Section 4 (Journeys), read the sentinel files specified in the gather table.
   • For --full mode: spawn Explore subagents for sections 1, 4, and 6. Subagents MUST run in foreground (do NOT use run_in_background). The briefing cannot render until all subagents return.
   • Health checks (Section 3) can run as background tasks since they're slow, but MUST complete before rendering.
4. WAIT GATE: Confirm ALL gather tasks (including subagents and background commands) have returned results before proceeding. Never render with assumed/guessed data.
5. Synthesize post-gather elements from gathered data:
   • Write the Headline (one sentence)
   • Extract Blockers from all sections (blocked lanes, failing checks, broken journeys)
   • Compute Delta if prior /status data exists in conversation
   • Compose Section 7: Daily Focus
6. Format the briefing in output order: Headline → Blockers → Last Session → Delta → Sections 1-6 → Daily Focus
7. Use freshness labels for all dates (today / yesterday / Nd ago / Nd ago ⚠ stale)
8. Output the complete briefing — one pass, no corrections afterward

---

Related Skills

• /architecture — Deep structural analysis (when status reveals architecture questions)
• /audit — Problem-finding mode (when status reveals health issues)
• /monitor — Continuous quality watching (when status shows failing checks)
• /plan — Feature planning (when status shows ready features to start)
• /debug — Investigation (when status shows broken journeys)