Claude Cockpit
tu11aa/claude-cockpit
Orquestração multi-projeto para agentes Claude em uma sessão de comando
Ver no GitHubOrquestrador de múltiplos projetos para agentes de código Claude. Controla capitães de projeto em sessões paralelas via cmux, com sincronização de status em Obsidian e suporte a comandos sob demanda para tarefas cross-project. Integra com Claude Memory, Task Master e GSD para fluxo de trabalho contínuo.
Instalação
npx convez-ai tu11aa/claude-cockpit1 comando no terminal: detecta o formato do repositório (marketplace ou plugin avulso) e instala via Claude Code. Requer Node 18+ e o Claude Code instalado.
Métricas
Casos de uso
- Gerenciar múltiplos projetos simultaneamente com agentes Claude dedicados
- Executar tarefas cross-project (briefing, revisão de aprendizados) sob demanda
- Sincronizar status de projetos para dashboard Obsidian e standups diários
- Coordenar crew de agentes com contexto fresco em ondas de execução
Veredito

Ferramenta especializada para fluxos multi-projeto complexos com Claude Code. Exige stack completo (cmux, Obsidian, plugins) e experiência com agentes — não é entrada para iniciantes, mas elegante para times que já operam com múltiplos projetos paralelos.
README
A tradução para o português deste README ainda está na fila.
claude-cockpit
Multi-project orchestration layer for coding agents. One command session controls everything.
Direction: Cockpit is moving to support all major coding agents — Claude Code, Codex, Cursor, Gemini CLI — not just Claude Code. Claude Code is the reference implementation today; other agents land through the plugin system's driver abstractions and the upcoming cross-agent projection layer (#31). See
docs/specs/2026-04-24-multi-agent-direction.md.
How It Works
cockpit launch <project> → Captain (per project, in cmux)
cockpit launch --all → Every Captain
cockpit command --task briefing → One-shot Command session in a split pane
(also: --task learnings-review | wiki-aggregate)
Captain → cockpit crew spawn … → Crew (new tab in the captain workspace, fresh agent CLI)
cockpit init— first-time setupcockpit launch <project>— start the project's captain in cmuxcockpit launch --all— start every captain at oncecockpit command --task briefing— on-demand Command session for cross-project work (optional; spawns in a split pane and exits when done)cockpit status— quick status check without spawning anything
Install
npm install -g claude-cockpit
cockpit init
cockpit doctor
Prerequisites
- Claude Code >= 2.1.32
- cmux (macOS terminal for coding agents)
- Obsidian (status tracking)
- Node.js >= 18
Required Integrations
# Claude Memory — cross-session continuity
/plugin marketplace add thedotmack/claude-mem
/plugin install claude-mem
# Task Master — PRD decomposition (works via Max subscription)
npm install -g task-master-ai
# GSD — wave-based execution for crew (fresh context per step)
npx get-shit-done-cc@latest --claude --global
See core/plugins.md for full plugin setup.
Obsidian Plugins
See obsidian/plugins.md for Dataview, Templater setup.
Commands
| Command | Description |
|---|---|
cockpit init | First-time setup — config, hub vault, scripts |
cockpit launch <project> | Start a specific project captain |
cockpit launch --all | Launch all captain workspaces |
cockpit command [--task <briefing|learnings-review|wiki-aggregate>] [--agent <a>] | Spawn a one-shot Command session in a split pane (no persistent Command). |
cockpit status | Show all project status (no Claude needed) |
cockpit standup | Daily standup summary (zero LLM tokens) |
cockpit doctor | Health check — verify dependencies |
cockpit projects list | List registered projects |
cockpit projects add <name> <path> | Register a project |
cockpit projects remove <name> | Unregister a project |
cockpit dashboard [--once] | Print a one-shot status grid for all projects to the terminal. |
cockpit dashboard --pane [--direction <dir>] [--interval <s>] | Open a refreshing sidebar pane in the current cmux workspace. |
cockpit dashboard sync-hub [--json] | Mirror spoke status.md files into {hubVault}/projects/ for Obsidian Dataview. |
cockpit runtime status <project> | Check if a project's captain workspace is running |
cockpit runtime send <project> <msg> | Send a message to a captain workspace (auto-Enter) |
cockpit runtime list | List all workspaces from the active runtime |
cockpit workspace read <project> <path> | Read a scope-relative file from the project's spoke vault |
cockpit workspace list <project> <dir> | List entries in a spoke vault directory |
cockpit workspace read --hub <path> | Read from the hub vault |
cockpit notify <message> | Send a message to the user via the configured notifier |
cockpit projection emit [--scope user|project] [--project <name>] [--target <name>] [--all] | Emit cockpit rules + skills to Cursor/Codex/Gemini config files |
cockpit projection diff [same flags] | Preview projection changes without writing |
cockpit projection list | Show registered projection targets and their destinations |
cockpit crew spawn <project> <task> [--name <n>] [--direction tab|right|left|up|down] [--agent <a>] | Spawn an interactive crew sub-session (tab in the captain workspace by default; --direction for a pane) |
cockpit crew send <project> <name> <message> | Send a follow-up turn to an existing crew |
cockpit crew read <project> <name> | Read a crew session's current screen |
cockpit crew close <project> <name> | Shutdown a crew session (closes its tab) |
cockpit crew list <project> | List live crews for a project |
cockpit shutdown [project] | Graceful shutdown |
cockpit feedback | Open opt-in feedback issue |
Architecture
Roles
- Command (Opus) — on-demand cross-project session. Spawned by
cockpit command --task <briefing|learnings-review|wiki-aggregate>in a split pane; exits when the task completes. No persistent Command process. - Captain (Opus) — project leader, uses Agent Teams + git worktrees
- Crew (Sonnet by default) — interactive sub-session running as a new tab in the captain's workspace (or a split pane via
--direction). Each crew is named (crew-1,crew-2, …) and stays idle between turns waiting for the captain's next message — same model as a Claude Agent Team subagent. Spawn withcockpit crew spawn, send follow-ups withcockpit crew send, close when done. Works with any agent CLI (claude and opencode are fully interactive; codex/gemini currently print-mode). Uses GSD for complex tasks.
Model Routing
Each role runs on the optimal model for cost/quality tradeoff. Configured in config.json:
- Command/Captain/Review: Opus (coordination + quality)
- Crew: Sonnet (execution)
- Exploration: Haiku (cheap lookups)
Runtime Abstraction
Workspaces run on a pluggable runtime driver (currently only cmux). Each project may override the global default via its runtime field. Bash scripts call cockpit runtime <op> to talk to the configured runtime instead of any specific binary. New runtimes (tmux, Docker, SSH) are added as driver files in src/runtimes/ — see docs/specs/2026-04-20-plugin-system-runtime-design.md.
Workspace Abstraction
Vault storage (hub + per-project spokes) runs behind a pluggable workspace driver (currently only obsidian). Filesystem operations — read, write, list, exists, mkdir — go through the driver instead of fs directly. Each project may override the global default via its workspace field. Bash scripts call cockpit workspace <op> to read/write vault data without hardcoding paths. New backends (Notion, plain-md, S3) are added as driver files in src/workspaces/ — see docs/specs/2026-04-21-plugin-system-workspace-design.md.
Notifier Abstraction
User-facing notifications run behind a pluggable notifier driver (currently only cmux). Escalations and other "tell the user" events go through cockpit notify <message>. The default CmuxNotifier delegates to cockpit runtime send --command — the abstraction exists as a swap-point for future Slack/Discord/email/pager drivers. Notifier is global (no per-project override). See docs/specs/2026-04-21-plugin-system-notifier-design.md.
Crew Spawn (Interactive Sub-Sessions)
Crew is the captain's equivalent of an Agent Team subagent — but runtime-agnostic. The captain spawns a crew via cockpit crew spawn <project> "<task>" [--name <n>], which opens a new tab in the captain's cmux workspace, boots an interactive Claude session (no -p), and sends the task as the first turn. The crew works on it and stays idle waiting for follow-ups. The captain drives the session with cockpit crew send/read/close/list, addressing each crew by its tab title (🔧 <project>:<name>).
Pass --direction right|left|up|down to use a split pane instead of a tab. State lives in the surface buffer + git; tabs die with the captain workspace on cockpit shutdown. Non-Claude agents (codex/gemini) currently still launch in print-mode; full interactive support is a follow-up. See docs/specs/2026-05-05-cockpit-thin-redirect-design.md.
Projection (Cross-Agent Config Sync)
Cockpit rules (Karpathy principles, captain-ops) and per-project AGENTS.md emit to each supported agent's canonical path via cockpit projection emit. User-level projection pushes cockpit's skills to ~/.cursor/rules/cockpit-global.mdc, ~/.codex/AGENTS.md, ~/.gemini/GEMINI.md. Project-level projection pushes a managed project's own AGENTS.md into {project}/CLAUDE.md, {project}/.cursor/rules/cockpit.mdc, {project}/GEMINI.md — zero cockpit-global content leaks into the project repo. Shared files use <!-- cockpit:start --> ... <!-- cockpit:end --> markers; dedicated files overwrite. See docs/specs/2026-04-24-plugin-system-projection-design.md.
The user-level projection now also inlines orchestrator/captain.generic.md and orchestrator/crew.generic.md as ## Captain Role / ## Crew Role sections inside the cockpit marker block, so non-Claude agents (Codex, Gemini, Cursor) load the same role descriptions Claude Code loads via --append-system-prompt-file. See docs/specs/2026-05-05-multi-agent-template-parity-plan.md (#45).
Obsidian Vaults (Hub-and-Spoke)
- Hub vault (
~/cockpit-hub) — cross-project dashboard + hub wiki - Spoke vaults — per-project status, learnings, and wiki
Knowledge System (opt-in writes)
- Status (opt-in) — captains record
{spokeVault}/status.mdviawrite-status.sh(also written by the captain session-end hook) when there's something worth noting (a blocker, "starting work on X"). Not on a schedule. - Dashboard —
cockpit dashboard --paneopens a refreshing sidebar pane in cmux that lists every project's live state, queried from the cockpit daemon's task records.cockpit dashboard sync-hubmirrors each spokestatus.mdinto{hubVault}/projects/so the hub vault'sdashboard.mdDataview query renders the same data inside Obsidian. - Handoff files — captain writes when in-flight work needs to survive into tomorrow; skipped on uneventful sessions.
- Daily logs — captain writes when the day produced something worth a log; not on a schedule.
- Learnings — recorded when a captain encounters a genuinely surprising or reusable pattern.
- Wiki — compiled, indexed knowledge pages in spoke vaults (
wiki/pages/); promoted from learnings when worth maintaining. - Hub Wiki — cross-project knowledge aggregated by an on-demand
cockpit command --task wiki-aggregaterun. - Scripts:
wiki-ingest.sh,wiki-query.sh,wiki-log.sh.
Session Continuity
- Handoff files — captain writes context on shutdown, reads on startup
- Session freshness — auto-detects new day or template changes, forces fresh context
- claude-mem — cross-session memory via MCP plugin
Config
~/.config/cockpit/config.json
{
"commandName": "command",
"hubVault": "~/cockpit-hub",
"runtime": "cmux",
"workspace": "obsidian",
"notifier": "cmux",
"projects": {
"brove": {
"path": "~/projects/brove",
"captainName": "brove-captain",
"spokeVault": "~/cockpit-hub/spokes/brove",
"host": "local",
"runtime": "cmux",
"workspace": "obsidian",
}
},
"defaults": {
"maxCrew": 5,
"worktreeDir": ".worktrees",
"teammateMode": "in-process",
"permissions": {
"command": "default",
"captain": "acceptEdits"
},
"models": {
"command": "opus",
"captain": "opus",
"crew": "sonnet",
"exploration": "haiku",
"review": "opus"
}
}
}
Supported Agents
| Agent | Status | Notes |
|---|---|---|
| Claude Code | ✅ Shipping | Reference implementation; reads CLAUDE.md, Skill tool, MCP via settings.json |
| Codex CLI | ✅ projection (skills + roles) | Captain/crew roles inlined into ~/.codex/AGENTS.md (#45). First-class role identity is #35. |
| Cursor | ✅ projection (skills + roles) | Captain/crew roles inlined into ~/.cursor/rules/cockpit-global.mdc (#45). |
| Gemini CLI | ✅ projection (skills + roles) | Captain/crew roles inlined into ~/.gemini/GEMINI.md (#45). |
| opencode | ✅ driver + projection (interactive crew) | opencode run "<prompt>" with --format json / -m <model>; AGENTS.md projects to ~/.config/opencode/AGENTS.md. |
Cross-agent config sync (one canonical source → agent-specific formats) is tracked in #31.
Inspirations
- Andrej Karpathy — coding principles baked into every captain/crew role (
plugin/skills/karpathy-principles/SKILL.md) - forrestchang/andrej-karpathy-skills — reference packaging for the four principles (MIT)
- Multica — validated the multi-agent runtime + skill-compounding direction
- AGENTS.md — convergence point for cross-agent instructions
- OpenSpace — self-improving learnings loop (record → capture → fix → mark-useful)
- ComposioHQ — tool/skill portability across agents
License
MIT