Captain Claw — Command a Fleet of AI Agents
Captain Claw is a multi-agent command center for spawning specialist teams, orchestrating workflows, and shipping software with deterministic Flows. It is self-hosted and MIT licensed.
About this tool
What is Captain Claw?
Captain Claw is a multi-agent command center (Flight Deck) that lets you command a fleet of AI agents. While most AI tools give you one agent in a chat box, Captain Claw lets you spawn specialist teams, run six orchestration modes, ship software with the Code pipeline, and compose deterministic Flows.
Core capabilities
- Multi-agent orchestration: Spawn specialist teams and run six orchestration modes.
- Deterministic Flows: Every Flow is a small text program, validated by a deterministic parser — not interpreted by an LLM reading YAML.
- Code pipeline: Ship software through decompose → review → execute → verify → re-plan, with an editable DAG you can inspect before anything runs.
- Agent Folders: Agent file ops never touch the host disk; all of a user's agents and sessions read and write the same tree, browsable in Flight Deck.
- Cost & trace analytics: Per-agent token and cost meters plus a full activity log with traceable lineage for every tool call, consult, and hand-off.
- Basna · Parallel ensemble: Specialists answer blind, then merge — N agents tackle the same question independently.
Deployment
Captain Claw is self-hosted, MIT licensed, and available on Linux, macOS, Windows, and Web.
Features & use cases
⇄ The handoff layer — backs inter-agent file transfer and the shared clipboard across the crew⊞ Shared across your fleet — all of a user's agents and sessions read and write the same tree⊟ Browsable in Flight Deck — explore, preview, and download files from the Agent Folders view1 Command a fleet of AI agents Most AI tools give you one agent in a chat boxA deterministic spine, agent judgment at the leaves Every Flow is a small text program, validated by a deterministic parser — not interpreted by an LLM reading YAMLA router sizes the job — small edits go straight to a specialistA stream of consciousness — timestamped thoughts and dreams, each spotting patterns across agentsagent file ops never touch the host diskAgent-written automations can never borrow your vetted authority until you approve themAgents query pointers, not source dumps, and stop re-reading your treeAI designs the team — roles, models, tools, and SOPs, all generated and editableBasna · Parallel ensemble Specialists answer blind, then merge N agents tackle the same question independently — no peeking at each other's work
Quick facts
Self-hosting supported
Open-source license: MIT