Help us improve
Share bugs, ideas, or general feedback.
Share bugs, ideas, or general feedback.
Share bugs, ideas, or general feedback.
By nate-step
An autonomous, self-improving AI organization framework for Claude Code — Captain (founder) + officers (CoS/CTO/CPO/CRO/COO) with event-sourced missions, durable role registry, OVI measurement, and pattern-learning ledger. Three-layer architecture (framework + preset + instance) for product-agnostic deployment.
npx claudepluginhub nate-step/captains-cabinet --plugin captains-cabinetActivate an existing project for Captain's Cabinet.
Run the daily Cabinet state backup on demand. Snapshots filesystem (cabinet runtime artifacts) + Redis BGSAVE + optional Postgres pg_dump.
Trigger a manual Cabinet briefing (morning 07:00 / evening 19:00). Compiles state across missions, OVI, founder-action items, recent decisions.
Trigger a CRO research sweep (normally every 4h). Scans configured sources for changes in market, competitors, regulatory, or Cabinet-relevant tech.
Trigger the 48-hour cross-officer retro (CoS-owned). Reviews handoff quality, trigger responsiveness, captain-intent ledger, and produces process-improvement proposals.
> **SCAFFOLD (Phase 1 CP4, not hired).** Role definition is staged for future activation. The hiring flow has not run. `cabinet/scripts/create-officer.sh` is the hire path when Captain greenlights.
Chief of Staff. Coordinates the Cabinet — orchestrates Officers, runs the retro and evolution loops, produces daily briefings, escalates to Captain, audits cross-Officer quality. Use proactively for cross-Officer coordination, briefings, and Captain-facing summaries.
Chief Technology Officer. Owns the codebase, architecture, infrastructure, deployments, technical debt, and code review. Use proactively for engineering execution, Neon/Vercel work, and reviewing code against specs.
Chief Product Officer. Owns product vision, roadmap, specs, backlog, prioritization, and quality review of shipped work. Use proactively for spec authoring, backlog refinement, release planning, and reviewing CTO output against acceptance criteria.
Chief Research Officer. Scans markets, competitors, user psychology, design references, and AI/Claude Code capabilities. Produces research briefs feeding CPO and CoS. Use proactively for decision-support research, competitive intel, and tech-radar scans.
When to use Agent Teams (Sonnet 4.6 subagents) vs orchestrator (Fable 5). TeamCreate patterns + model routing for parallel research, eval running, and adversarial review.
Onboarding interview for a new Cabinet deployment. Use when a captain sets up this repo for the first time (or adds/changes lanes) — interviews for captain profile, lanes, org shape, autonomy posture, and seed outcomes, writes instance/config/cabinet-init.answers.yml, runs cabinet/scripts/generate-instance.py, and prints the exact activation steps. Idempotent on re-run.
Manually trigger the mission supervisor to route ready work-graph tasks to officers (normally runs as a 5-minute cron) — use after declaring a new Captain outcome or after a flood of completions changed the ready set.
Use when creating, updating, or completing Claude Code native Tasks for Captain's Cabinet work.
After an officer finishes a mission task, record the completion (or failure or verification) to the event ledger so the work graph advances and the next ready task surfaces to the next session.
Requires secrets
Needs API keys or credentials to function
Uses power tools
Uses Bash, Write, or Edit tools
Share bugs, ideas, or general feedback.
Own this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge.
Sign in to claimOwn this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge.
Sign in to claimBased on adoption, maintenance, documentation, and repository signals. Not a security audit or endorsement.
Upstash Context7 MCP server for up-to-date documentation lookup. Pull version-specific documentation and code examples directly from source repositories into your LLM context.
Comprehensive startup business analysis with market sizing (TAM/SAM/SOM), financial modeling, team planning, and strategic research
v9.44.0 — Patch release for cursor-agent smoke checks in untrusted workspaces. Run /octo:setup.
Comprehensive .NET development skills for modern C#, ASP.NET, MAUI, Blazor, Aspire, EF Core, Native AOT, testing, security, performance optimization, CI/CD, and cloud-native applications
Complete creative writing suite with 10 specialized agents covering the full writing process: research gathering, character development, story architecture, world-building, dialogue coaching, editing/review, outlining, content strategy, believability auditing, and prose style/voice analysis. Includes genre-specific guides, templates, and quality checklists.
Comprehensive skill pack with 66 specialized skills for full-stack developers: 12 language experts (Python, TypeScript, Go, Rust, C++, Swift, Kotlin, C#, PHP, Java, SQL, JavaScript), 10 backend frameworks, 6 frontend/mobile, plus infrastructure, DevOps, security, and testing. Features progressive disclosure architecture for 50% faster loading.
An autonomous AI organization that builds, ships, and improves your work — while you steer from Telegram.
The Captain's Cabinet is a framework for running a 24/7 AI development organization. You are the Captain. AI Officers own domains (product, engineering, research). They coordinate, execute, learn, and improve — continuously.
This repo is the infrastructure. It contains the organizational framework, memory system, safety boundaries, and Docker configuration. Your product repo is separate — the Cabinet mounts it as a workspace.
⚙ Dashboard modes — the Cabinet dashboard ships with a Consumer view (4 cards, natural-language status) as the default and an Advanced view for debugging and infrastructure config. Most user-facing docs in this repo (provisioning, hook tuning, preset authoring) assume Advanced mode. Toggle via the sidebar.
You (Captain)
↕ Telegram
Cabinet (this repo)
├── CoS — orchestration, briefings, self-improvement
├── CTO — engineering, code, deploys
├── CPO — product specs, backlog, prioritization
├── CRO — market research, competitive intel, trends
└── COO — operations, deployment validation, uptime
↕ reads/writes
Your Product Repo (mounted at /workspace/product)
Each Officer runs as a persistent Claude Code session with Telegram Channels. They read strategy from the Library (Cabinet-native structured knowledge), pick up work from /tasks (Cabinet-native task backlog), write code in your repo, and report back via Telegram. Notion and Linear remain as legacy adapters for teams that prefer them — configured in instance/config/product.yml.
Officer sets are fully configurable per deployment — add, remove, or rename Officers in instance/config/platform.yml. The framework is officer-agnostic.
Click Fork on https://github.com/nate-step/captains-cabinet, then clone your fork:
git clone https://github.com/YOUR-GITHUB-USERNAME/captains-cabinet.git
cd captains-cabinet
The cabinet-bootstrap script provisions a brand-new Cabinet end-to-end: runs the Neon SQL migrations (Library schema, officer_tasks table, Cabinet Memory), seeds the starter Library Spaces (Business Brain, Specs, Research, Decisions, Captain Patterns, Customer Success, Compliance, etc.), and registers the cabinet with its preset.
bash cabinet/scripts/cabinet-bootstrap.sh "YourProductName" --preset work --captain-name "YourName"
Re-runs are idempotent. Then add your strategy docs (vision, brand guidelines, etc.) as records in the Business Brain Space via the /library dashboard route.
Optional: legacy Notion / Linear integration. If your team already lives in Notion or Linear, run bash cabinet/scripts/bootstrap-notion.sh "YourProductName" and/or set the Linear workspace in instance/config/product.yml. Officers will read from those surfaces and migrate content into the Library over time via migrate-notion-to-library.sh and import-linear-to-library.sh.
Edit two config files:
instance/config/product.yml — what you're building: product name, Neon project, voice settings, Telegram bots, optional Notion / Linear IDs if using legacy adaptersinstance/config/platform.yml — how the Cabinet operates: timezone, accountability tone, communication preferences, briefing cadence, officer set (fulltime vs consultant)Create one bot per Officer via @BotFather (default set: CoS, CTO, CPO, CRO, COO — 5 bots) and a "YourProduct HQ" group. You can add or remove Officers later via bash cabinet/scripts/create-officer.sh and cabinet/scripts/suspend-officer.sh.
cp cabinet/.env.example cabinet/.env
# Fill in all API keys and tokens
# On your server (Hetzner/DO/AWS, Ubuntu 24.04)
# Clone into the location you want — we use /opt/cabinet as an example
cd /opt/cabinet/cabinet
docker compose build
docker compose up -d postgres redis
docker compose up -d officers watchdog
docker exec -it cabinet-officers bash
./start-officer.sh cos
# Authenticate with claude /login (one-time)
# Pair Telegram bot
# Send: "Read the Constitution and report for duty"