# Crewship: All hands on your deck.

> Welcome aboard. Crewship is an Apache-2.0, open-source, self-hosted platform for AI agent crews. Each crew runs in its own Docker (or Podman / Apple Containers) boundary. Agents can run any official CLI — mix Claude Code, Codex, Gemini, and OpenCode in the same crew, each running the model best suited to their role. Current version: v0.1.0-beta.4.

## What makes Crewship different

Crewship is neither a cloud-only vendor tool (like Claude Code, Codex, or Devin) nor a Python library (like CrewAI, LangGraph, or AutoGen). It is a complete self-hosted product that combines three things no one else has together:

- **Container isolation per crew** — one Docker container per team of agents, non-root UID, sidecar architecture, prompt-injection attempts stop at the container boundary
- **Multi-agent hierarchy** — LEAD orchestrates, AGENT executes, COORDINATOR aligns crews across workspaces. Hierarchy is functional (who orchestrates), not value-based (who matters)
- **CLI neutrality** — Claude Code, Codex CLI, Gemini CLI, OpenCode and any OpenAI-compatible endpoint, all in one crew with one journal and one budget

## Core modules

Every surface in Crewship is a read-model over a single append-only event stream, the Crew Journal.

- **Crew Journal** — canonical source of truth for every agent action; append-only, typed, queryable, retention-aware
- **Paymaster** — hierarchical LLM budgets (workspace → crew → mission → agent) with hard stops before the call is made
- **Keeper** — AES-256-GCM credential vault; agents request access, an LLM gatekeeper audits every grant
- **Harbormaster** — first-class human-in-the-loop approval queue; sync or async gates before high-risk actions
- **Cartographer** — named mission checkpoints with fork and restore over the journal cursor
- **Crow's Nest** — live terminal, network, filesystem, and resource view for every crew (owner/admin only)
- **Lookout** — guardrails: prompt-injection detection, schema validation, secret redaction
- **Watch Roster** — real-time agent presence across every crew in the workspace
- **Fleet** — workspace-level overview of all crews and agents, their status, cost, and activity

## Install

One command. Requires a container runtime (Docker Desktop, Podman Desktop, or Apple Containers on macOS 26+).

- macOS: `brew install crewship`
- Linux: `curl -fsSL https://crewship.ai/install | bash`
- Windows: `docker run -p 8080:8080 crewship/crewship:latest`

A crew is running in 60 seconds.

## Stack

Single 30 MB Go binary with an embedded Next.js 15 UI. PostgreSQL 16 (or SQLite for development). Docker-per-crew isolation. AES-256-GCM encryption at rest. Apache-2.0 licensed. Works on macOS, Linux, and Windows (via Docker or WSL2).

## Philosophy

Agents are persistent colleagues, not tools. Each has a role, credentials, memory, and a place in the hierarchy — LEAD, AGENT, or COORDINATOR — because that is how real work gets done. Every action lands in one append-only stream; observability is the substrate, not an afterthought tacked on later.

## Audience

Built for people who run their own stack:
- Founders who want an AI crew that plans, codes, reviews, and ships with budgets they actually control
- Solo builders who refuse vendor lock-in, per-seat pricing, or "contact us" pages
- Open-source maintainers who want something they can fork, self-host, and extend

## Links

- Homepage: https://crewship.ai
- Full content for LLMs: https://crewship.ai/llms-full.txt
- Documentation: https://docs.crewship.ai (LLM-optimized: https://docs.crewship.ai/llms-full.txt)
- GitHub repository: https://github.com/crewship-ai/crewship
- Releases: https://github.com/crewship-ai/crewship/releases
- Community Discussions: https://github.com/crewship-ai/crewship/discussions
- Discord: https://discord.gg/crewship

## Sections on the homepage (in scroll order)

### Explore block — what it is, how it works, who it's for
- Live demo: https://crewship.ai/#demo
- Crew Journal detail: https://crewship.ai/#journal
- Crow's Nest detail: https://crewship.ai/#crowsnest
- Paymaster detail: https://crewship.ai/#paymaster
- Features: https://crewship.ai/#features
- How it works: https://crewship.ai/#how
- Use cases: https://crewship.ai/#usecases

### Ship log block — trust, proof, momentum
- Compare with alternatives: https://crewship.ai/#compare
- Philosophy: https://crewship.ai/#philosophy
- Changelog + roadmap voting: https://crewship.ai/#changelog
- FAQ: https://crewship.ai/#faq