Skip to content

Destynova2/cli-code-skills / cli-forge-chef

cli-forge-chef

Unverified

Generate a multi-agent orchestrator for any project using the Brigade de Cuisine pattern: Chef (plans, decides), Sous-Chef (validates, merges), Commis (execute). Clients order via a Menu (PERT), the kitchen delivers Plats (merged features). Creates tmuxinator config, prompts, shared-state memory, quality gates, and permissions. Uses Claude Code Agent Teams (TeamCreate + SendMessage) with visible tmux panes. Use when the user wants to parallelize work across multiple Claude Code agents, orchestrate a multi-agent sprint, or says 'boss', 'multi-agent', 'orchestrate', 'parallelize agents', 'team of claudes', 'sprint multi-agent', 'brigade', 'chef'. Also triggers on 'swarm', 'agent team', 'parallel workers', 'PERT multi-agent'.

references/shared-state-template.md

View on GitHub
# Template — Shared State

Generate this file at `{project}/.claude/shared-state.md`.
Replace `{variables}` with the project values.

---

```markdown
# {Project Name} Shared State — Session {session_name}

> Shared memory between all parallel Claude Code sessions.
> **CHEF**: only agent allowed to write in "Valid merges" and "Green light".
> **COMMIS**: read this file BEFORE coding. Write in "In progress" and "Done".
> **EVERYONE**: re-read this file after every merge to see what changed.

## Green light — Resolved dependencies

> The Chef writes here when a merge is validated and dependents can continue.
> Commis: if your task depends on another, wait until it appears here.

| Merged branch | Tests | CI | Date | Unblocks |
|---------------|-------|----|------|----------|

## In progress

| Worktree | Branch | Task | Files touched | Started |
|----------|--------|------|----------------|---------|

## Done (awaiting merge)

| Worktree | Branch | Task | Result | Tests | Files modified | Date |
|----------|--------|------|--------|-------|-----------------|------|

## Maître d'hôtel surveillance (in-flight PRs post-merge)

> The Maître d'hôtel owns this section. Sous-Chef appends a new row when handing off a PR;
> M'H updates Status on every 45 s poll; M'H moves the row to "Valid merges" on Encaissement.
> Any PR here is NOT considered merged — "auto-merge enabled" is not "merged".

| PR | Branch | Sent by | Status | Last check | Relaunches | Issue |
|----|--------|---------|--------|------------|------------|-------|

**Status legend:** En salle | Surveillance | Rattrapage | Relance | Renvoi | Escalade | Encaissement

## Valid merges

| Branch | Merge commit | CI run | Status | Date | Tag |
|--------|-------------|--------|--------|------|-----|

## Potential conflicts

> If you edit a file already listed in another worktree's "In progress", note it here.

| File | Affected worktrees | Risk | Resolution |
|------|--------------------|------|------------|

## Decisions made

> Technical choices made during the sprint that affect other sessions.
> Commis: read this section BEFORE coding, a choice made by another worker may change your approach.

| Decision | Reason | Impacts | By | Date |
|----------|--------|---------|----|------|

## Quality Gates — Metrics (filled by the Chef)

> The Chef fills this section after every audit. Commis: if a gate fails, fix before resubmitting.

| Gate | Scope | Score | Threshold | Status | Date |
|------|-------|-------|-----------|--------|------|
{quality_gates_rows}

### Strong couplings (from /cli-audit-tangle)

> Chef fills this before assigning tasks. Commis: do NOT touch coupled files assigned to another worker.

| Module/Function | Coupling | Assigned to |
|------------------|----------|-------------|

## PERT (computed in Phase 0.5 — see references/pert-computation.md)

> Scheduling contract for the sprint. Chef writes once at Phase 0.5.
> Recomputed only on apoptosis, DENY past P, or user-added plat — never on every merge.
> **Format is a Mermaid `flowchart LR` — never a `gantt`**. A PERT is a dependency DAG, not a timeline.

```mermaid
flowchart LR
    start([Sprint start])
    %% plats go here, critical-path nodes get :::critical
    done([Envoye])
    classDef critical stroke:#E74C3C,stroke-width:3px,fill:#FDEDEC
```

| Plat | O | M | P | E | σ | ES | EF | LS | LF | slack | critical |
|------|---|---|---|---|---|----|----|----|----|-------|----------|

**Makespan:** {makespan} commis-hours on the critical path.
**95% CI:** {makespan} ± {ci} commis-hours.
**Critical path:** {critical_path}.

### Scheduled view (derived from PERT — wall-clock projection)

> Post-scheduling `gantt` derived from ES/EF of the PERT. NOT the PERT itself.
> The PERT above is the dependency contract; this gantt answers "when will it be done?".
> Regenerated only when the PERT is recomputed (apoptosis / DENY past P / new plat).

```mermaid
gantt
    title Sprint schedule (post-PERT projection)
    dateFormat X
    axisFormat %Hh
    section Critical path
    auth-refactor   :crit, 0, 4h
    api-endpoints   :crit, after auth-refactor, 3h
    section Buffer
    schema-migrate  :2h, 0, 2h
    docs-update     :1h, after api-endpoints, 1h
```

Critical-path tasks use `crit`. The section split makes the slack-free chain visually separable from the buffered tasks.

## Task pool (dispatch queue — consumed by free commis)

> Two views of the same state: a `kanban` for one-glance status, a table for machine parsing.
> Derived from the PERT. `Ready = 1` iff all predecessors are Envoye.
> Dispatch rule: highest Priority (longest path to done) wins, ties broken by smaller slack,
> then smaller E. File-exclusion: never dispatch two rows whose Write-sets intersect.

```mermaid
---
config:
  kanban:
    ticketBaseUrl: ''
---
kanban
    Ready[Ready — pickable]
        auth-refactor[auth-refactor<br/>pri=12 slack=0]@{ priority: 'High' }
        schema-migrate[schema-migrate<br/>pri=8 slack=2]@{ priority: 'Medium' }
    InProgress[In progress — locked to a commis]
    Blocked[Waiting on predecessor]
        api-endpoints[api-endpoints<br/>pri=8 slack=0<br/>waits: auth-refactor]@{ priority: 'High' }
        docs-update[docs-update<br/>pri=1 slack=4<br/>waits: api-endpoints]@{ priority: 'Low' }
    Envoye[Envoye — merged + CI green]
```

**Columns:** `Ready` (pickable now) / `InProgress` (a commis is on it) / `Blocked` (waiting on a predecessor) / `Envoye` (merged). A plat moves Ready → InProgress on dispatch, InProgress → Envoye on merge. A Blocked plat jumps to Ready when its last predecessor becomes Envoye.

**Priority label** (`High` / `Medium` / `Low`) is derived from the PERT priority and maps directly to dispatch order. Critical-path plats (`slack = 0`) are always `High`.

| Plat | Ready | Priority | Slack | Write-set | Commis |
|------|-------|----------|-------|-----------|--------|

## Sensitive zones (3/3 unanimity quorum)

> Reviewed at the end of every sprint. If a file causes problems, move it here.
> If a file stays here for 3 sprints with no incident, demote it back to the normal zone.
>
> **Machine-parseable format**: the authoritative list is the `BOSS_SENSITIVE_PATHS`
> block below. The human table is generated from that block and serves as documentation.
> The Sous-Chef `/loop` parses only the block — always keep it up to date.

<!-- BOSS_SENSITIVE_PATHS:START -->
```sensitive-paths
# One glob pattern per line. Lines starting with # are comments.
# Parsed by /loop Sous-Chef and the Chef shutdown protocol.
# Format: <glob>    # <reason>
.github/workflows/**          # CI = global impact
**/Cargo.toml                 # Supply chain risk (Rust deps)
**/package.json               # Supply chain risk (npm deps)
**/go.mod                     # Supply chain risk (Go deps)
**/pyproject.toml             # Supply chain risk (Python deps)
**/requirements*.txt          # Supply chain risk (Python deps)
.env                          # Secrets
**/*.secret                   # Secrets
**/credentials*               # Secrets
src/auth/**                   # Critical authentication
src/security/**               # Critical security
CONTRACTS.md                  # Project rules
CONTRIBUTING.md               # Project instructions
```
<!-- BOSS_SENSITIVE_PATHS:END -->

### Human table (generated from the block above)

| Pattern | Reason | Since | Incidents |
|---------|--------|-------|-----------|
| .github/workflows/** | CI = global impact | initial | - |
| **/Cargo.toml | Supply chain risk (Rust) | initial | - |
| **/package.json | Supply chain risk (npm) | initial | - |
| .env, **/*.secret, **/credentials* | Secrets | initial | - |
| src/auth/** | Critical authentication module | initial | - |
| src/security/** | Critical security module | initial | - |
| CONTRACTS.md, CONTRIBUTING.md | Project rules | initial | - |

### How to parse the block (for /loop or scripts)

```bash
# Extract pure patterns (no comments, no markers)
sed -n '/<!-- BOSS_SENSITIVE_PATHS:START -->/,/<!-- BOSS_SENSITIVE_PATHS:END -->/p' \
  {project}/.claude/shared-state.md \
  | sed -n '/```sensitive-paths/,/```/p' \
  | grep -v '^```' \
  | grep -v '^#' \
  | grep -v '^$' \
  | awk '{print $1}'
```

### Hallucination history

> If a Sous-Chef APPROVEd a diff that caused a problem, record it here.
> Used to improve the Sous-Chefs' rules in the next sprint.

| Sprint | Sous-Chef | File | What happened | Fix applied |
|--------|-----------|------|----------------|-------------|

## Shared context

> Information every worker must know.

{context_items}
```
reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis.Stars: 467kUpdated Claude CodeCodexVerifiedInstall
diagnosing-superpowersobra/superpowersUse when a superpowers session went wrong and your human partner wants to know why — repeated work, ignored plans, stumbles, poor results, a skill that didn't fire, "it took too long", "why is it so expensive", "what is it doing" — or wants to build a bug report for the superpowers maintainers, for the current session or a past one identified by id or path, on any harness.Stars: 319kUpdated Claude CodeCodexVerifiedInstall
dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependenciesStars: 319kUpdated Claude CodeCodexVerifiedInstall
executing-plansobra/superpowersUse when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is availableStars: 319kUpdated Claude CodeCodexVerifiedInstall