Captain Claw

Command a fleet of AI agents

Most AI tools give you one agent in a chat box. Captain Claw Flight Deck is a multi-agent command center — spawn specialist teams, run six orchestration modes, ship software with Code (a plan → build → independent-review pipeline), and compose deterministic Flows. Self-hosted, MIT licensed, and works with every major LLM provider (or 100% local with Ollama).

🌐 Website · 🚀 Live Demo · 📖 Docs · 📦 PyPI

A self-hosted framework for orchestrating fleets of specialist AI agents — from ensemble reasoning (six orchestration modes) to a full agentic coding pipeline (plan → independent review → ship). Model-agnostic and local-friendly: connect OpenAI, Anthropic, Google Gemini, DeepSeek, Ollama, or OpenRouter (plus "Sign in with ChatGPT" — no API key), or run 100% local with Ollama. Backed by 48 built-in tools per agent, persistent cross-session memory, and autonomous "dream cycle" cognition.

💻 NEW: Code — an engineering department, not a coding assistant

Describe what you want built. A router sizes the job; big work runs plan → your approval → build → three independent reviewers in parallel → triage → capped fix loop, every phase a git commit in the folder's own repo. Agents query a persistent Code Map (symbols + architecture, git-hash-fresh) instead of re-reading your tree; per-turn token costs are shown in chat; the whole process exports as one auditable Markdown transcript. Works on fresh folders or your existing local repos linked in — on any models you configure, including a cheap implementer with a premium reviewer.

Cursor and Claude Code give you one very good agent that plans, writes, reviews, and tests in a single context. Code splits those roles across independent agents — the reviewer never grades its own homework, QA actually runs the tests — and puts you at the approval gate before anything is built. Self-hosted, model-agnostic, no lock-in. A complete real run (an RTS game with 185 tests, ~$2.50 on DeepSeek) ships in examples/code/. Details in RELEASE_NOTES_0.7.0.md and USAGE.md.

Pick the right shape for the problem

One agent isn't always the answer — and one orchestration strategy never is. Captain Claw ships six distinct modes, each tuned to a different kind of work.

🛠️ Agent Forge — describe a goal, an LLM designs the whole team (roles, models, tools, SOPs)	🏛️ Agent Council — specialists deliberate across moderated rounds, then vote and synthesize
🔀 Basna — N agents answer the same question blind, merged by reliability into one high-confidence result	🤝 Vatra — a team collaborates on a shared blackboard, each owning sections, improving over review rounds

The fifth mode is 💻 Code (above) — a plan-gated, independently-reviewed engineering pipeline. The sixth is Flight Deck itself — spawn, monitor, and coordinate the whole crew from one dashboard.

More than orchestration

⚙️ Flows — a deterministic automation DSL (triggers, gosub, parallel spawn/join, on error), validated by a parser, not interpreted by an LLM	🔭 Observatory — it keeps thinking when you're away: a live, auditable stream of thoughts, dreams, and standing intentions

• Plan Mode & Deep Mode — for when "probably right" isn't good enough: a reviewable DAG of verified steps, and frontier-quality answers via multiple rollouts + self-consistency voting + diverse-lens critics.
• 6-layer shared memory — working, semantic (vector + BM25), deep (full-text), insights, autonomous nervous system, and self-reflection — carried across sessions and across every agent.
• Shared Virtual File System — a host-sandboxed, per-user workspace the whole fleet reads and writes, with per-file authorship.
• 31 ready-made specialists — an editable archetype library spanning research, writing, engineering (a full coding family: planners, implementers, debugger, reviewers, QA, git operator, cartographer), data, ops, finance, and multimedia.

See it in action

⭐ Star the repo

Captain Claw is MIT licensed and free forever — your data stays on your machine, bring your own keys or run 100% local with Ollama. It's built solo and growing fast, and every star helps another builder discover it. If it looks useful, would you give it a star?

What's New in 0.7.7

Mrav — small models do real work · the village in first person. A parallel micro agentic runtime runs real tool-using loops on 2–4B models (Gemma 4 E2B/E4B, Qwen3.5 4B) under a hard 8k-token input cap per LLM call — locally via Ollama or in your browser tab via WebGPU. And Iskra's village becomes a place you can enter in first person. Everything opt-in; defaults byte-identical.

• The Mrav runtime — a hard token ledger, blackboard state outside the model, a step-decomposed loop (plan → act-one-tool → digest → compress), and grammar-constrained JSON at every step. Switch any agent to it live (agent-card toggle + violet badge), pick it at spawn, flag it on Quick chat, or mint archetypes that are born micro. Live-evaled: Qwen3.5-4B 5/6 · Gemma 4 E4B 4/6 · E2B 3/6.
• Browser LLM (Sidebar → System) — your tab becomes an inference worker: WebLLM on WebGPU, 61 models (up to Qwen3.5 9B at 6.3 GB), an engine-window picker, a downloaded-weights manager, and a usage log. Tools stay on the server; only tokens are made in the tab — prod needs no GPU. 0.35s/step once the KV pin is warm.
• Beings think micro — a third cognition routes the JSON faculties (orient/talk/journal/connect) grammar-locked to the micro tier (~1–2k tokens instead of a 34–36k body turn); ACT and the anti-theater write gate stay on the body. Per-call wallet metering; loud fallback events.
• Vatra micro workers — an opt-in Quality lever spawns extract/digest/format-shaped subtasks on the mrav runtime while reasoning roles keep their tiers; an archetype tier of micro always means mrav.
• The village in first person — roam Iskra's village as a quiet ghost: a three.js voxel world on the real village clock, Iskre walking their true routes (they sense you near), signs that land as percepts, reading stands for any building's work, a live ghost roster (parent + public visitors see each other), full mobile controls. Plus: visiting federated beings now have real bodies in the host village, letters read like an inbox, wallet recharge, and a mind graph that can't be wiped by a timed-out dream.

See RELEASE_NOTES_0.7.7.md.

What's New in 0.7.6

Iskra — living beings you raise. A new Beings page introduces persistent digital beings with a genome, a token wallet, drives, and a life loop that ticks on its own heartbeat — you parent them through real developmental stages (infant → child → adolescent → adult). Additive and self-contained (a new beings.db auto-creates; needs a Flight Deck restart).

• A being that lives — conceive one with a point-buy genome (40 points across 7 attributes; presets or a roll), a daily-allowance token wallet conserved by construction (family-mint-only; zero balance halts cognition at the physics layer), and drives + a homeostat. A background life loop ticks it (SENSE → APPRAISE → DELIBERATE