# Crewship — Full Content for LLMs > Complete marketing-site content, machine-readable. Live data refreshed hourly from GitHub (latest release published 2026-05-19). Current version: v0.1.0-beta.4. ## Positioning **Brand line (always-on):** Welcome aboard. **Headline:** All hands on your deck. **Subheadline:** Crewship is the self-hosted platform for AI agent crews. Mix Claude Code, Codex, Gemini, or OpenCode in one crew. Docker-isolated. Journal-backed. Everything on your side. **Multi-CLI design:** Each agent in a crew can run a different official CLI — Claude Code (Anthropic), Codex (OpenAI), Gemini CLI (Google), OpenCode (open-source). Mixed within the same crew via failover pools. This is a first-class feature, not a checkbox. Native Ollama-as-runtime support is on the roadmap. ## Three pillars (what no one else has together) ### Container isolation per crew *The ship's hull — each crew lives in its own Docker container.* One container per crew (crewship-team-{slug}). Non-root UID 1001. Sidecar UID 1002. cap-drop ALL. Configurable network policy. Prompt injection stops at the container boundary. ### Mix any AI agent *Each agent in a crew can run a different provider. No vendor lock. No single point of failure.* CLI-neutral adapter layer. Claude Code for implementation, Gemini for research, OpenCode (open-source) for sensitive work — all in the same crew. Failover pools keep the crew sailing if one provider goes down. ### Multi-agent hierarchy *LEAD orchestrates. AGENT executes. COORDINATOR aligns crews.* LEAD (one per crew) delegates via HTTP sidecar API. COORDINATOR (one per workspace) orchestrates across crews. Hierarchy is functional (who orchestrates), not value-based (who matters). ## Trust & control ### Paymaster *The accountant who stops you before you overspend.* Hierarchical budgets: workspace → crew → mission → agent. Hard stops, not surprises. ### Keeper *A vault with a gatekeeper who reads the request.* AES-256-GCM credential store. Sidecar injects at request-time; agent never sees plaintext. LLM gatekeeper logs every grant. ### Harbormaster *The gate where humans approve before agents act.* HITL approval queue. Sync or async. Timeouts. Full history. First-class, not retrofit. ## Platform - **Crew Journal** — Append-only event stream. Every feature is a read-model over it. - **Crow's Nest** — Live terminal, network, filesystem, resources per crew. Admin only. - **Single binary** — 30 MB Go binary. Embedded SQLite + Next.js. No Node.js at runtime. Also in the box: Cartographer (checkpoints & fork), Lookout (prompt-injection guardrails), Watch Roster (agent presence), Fleet (workspace overview), Quartermaster (evals). ## Works with - Claude Code (Anthropic) - Codex CLI (OpenAI) - Gemini CLI (Google) - OpenCode (open-source) - Any OpenAI-compatible endpoint - Ollama as agent runtime — roadmap One crew, one journal, one budget — regardless of which provider each agent uses. ## Install One command. Requires Docker Desktop, Podman Desktop, or Apple Containers on macOS 26+. ### macOS ``` brew install crewship # or docker run -p 8080:8080 crewship/crewship:latest ``` ### Linux (Debian, Ubuntu, Fedora, Arch) ``` curl -fsSL https://crewship.ai/install | bash # or docker run -p 8080:8080 crewship/crewship:latest ``` ### Windows ``` docker run -p 8080:8080 crewship/crewship:latest # or inside WSL2: curl -fsSL https://crewship.ai/install | bash ``` A crew is running in 60 seconds. ## How it works 1. **Install** — 30 MB Go binary, embedded SQLite for development, optional PostgreSQL for production. No Node.js at runtime. 2. **Create a crew** — pick a template (Development, Research, Support, Marketing) or build your own. Each crew becomes an isolated Docker container. 3. **Run missions** — delegate work to agents. Every action lands in the journal, every budget is enforced, every approval is logged. Stack: Go backend · Next.js 16 · PostgreSQL 16 · Docker isolation · AES-256-GCM. ## Comparison with alternatives Crewship sits in the gap between "one cloud agent" and "build it yourself with CrewAI". It is neither a cloud-only product nor a Python library — it is a self-hosted platform. | Feature | Crewship | Claude Code | CrewAI | AutoGen | |---|---|---|---|---| | Multi-agent crews | ✓ LEAD + agents | ✗ Single agent | ✓ Crew abstraction | ✓ Conversational | | Drop-in self-hosted runtime | ✓ One Go binary | ✗ Vendor cloud | ~ Library, you wire | ~ Library, you wire | | Mix CLIs in one crew | ✓ Claude · GPT · Gemini · Ollama | ✗ Anthropic only | ~ API wrappers, manual | ~ API wrappers, manual | | Uses official vendor CLIs | ✓ Subscription-friendly | ✓ Own CLI only | ✗ Direct API | ✗ Direct API | | Container isolation per agent | ✓ Docker / Podman | Vendor VM | ✗ Same process | ✗ Same process | | Human-in-the-loop approvals | ✓ Harbormaster gate | ~ Per-tool prompt | ✗ DIY | ✗ DIY | | Hierarchical LLM budgets | ✓ Workspace → agent | ~ Per-user cap | ✗ — | ✗ — | | Audit trail (typed journal) | ✓ 25+ entry kinds | ~ Session log | ✗ BYO telemetry | ✗ BYO telemetry | | Live observability UI | ✓ Crow's Nest | ✗ — | ~ LangSmith addon | ✗ — | | License | Apache-2.0 | Proprietary | MIT | Creative Commons | Not competitors we're trashing — tools we're complementing. Crewship runs Claude Code inside a crew. ## Use cases - **For founders** — stop gluing scripts together. Run a crew that plans, codes, reviews, and ships, with budgets you actually control. - **For solo builders** — your laptop, your agents, your rules. No vendor lock. No per-seat pricing. No "contact us." - **For open-source maintainers** — fork it, self-host it, extend it. Apache-2.0. CLI-first. Hackable all the way down. ## Philosophy Crewship treats agents like hires, not function calls. Each has a role, credentials, memory, and a place in the hierarchy — LEAD, AGENT, COORDINATOR — because that is how real work gets done. Hierarchy here is **functional** (who orchestrates), not **value-based** (who matters). A well-run crew of AI colleagues is a better metaphor than a smarter chatbot. Every action lands in one append-only stream. The journal is not observability tacked on — it is the substrate. ## Recent releases ### v0.1.0-beta.4 · 2026-05-19 · prerelease **Crewship v0.1.0-beta.4** Install with Homebrew: Or download the binary for your platform below. All release artifacts are signed with Sigstore cosign (keyless, GitHub Actions OIDC) and ship with SBOMs in both SPDX and CycloneDX formats. Verify a binary signature: f… [Release notes](https://github.com/crewship-ai/crewship/releases/tag/v0.1.0-beta.4) ### v0.1.0-beta.2 · 2026-05-18 · prerelease **Crewship v0.1.0-beta.2** Install with Homebrew: Or download the binary for your platform below. All release artifacts are signed with Sigstore cosign (keyless, GitHub Actions OIDC) and ship with SBOMs in both SPDX and CycloneDX formats. Verify a binary signature: e… [Release notes](https://github.com/crewship-ai/crewship/releases/tag/v0.1.0-beta.2) ## Roadmap (voting on GitHub Discussions) - **Kubernetes container provider** [planned] · 87 upvotes — Run crews as K8s pods. Horizontal scale, namespace per workspace, standard RBAC integration. - **MCP bridge for tools** [considering] · 64 upvotes — Expose any MCP server to all agents in a crew — gated by Keeper, logged in Journal. - **Public skills marketplace** [considering] · 53 upvotes — Curated, sandbox-enforced skills. Every install is reviewed and runs inside crew boundary. ## FAQ ### Can I mix different AI providers in the same crew? Yes — this is the core of how Crewship works. Each agent in a crew can run a different CLI. For example: your LEAD agent runs on Claude Code for orchestration, a researcher agent runs on Gemini for its long context window, a coding agent runs on Codex, and a sensitive-data scout runs on OpenCode (open-source CLI). Failover pools mean if one provider goes down, the crew keeps working. ### Will there be a hosted cloud version of Crewship? Not in the short term. Self-hosting is not a pricing tier for us — it is the identity of the product. Crewship exists because AI tools that run in someone else's cloud, with someone else's credentials, are not really yours. A managed hosted option may eventually appear alongside the open-source core (Apache-2.0), but it will always be optional. The self-host path will never be deprecated. ### Do I need Docker to run Crewship? Yes — Crewship runs every crew in an isolated container. You need either Docker Desktop (macOS/Windows), Docker Engine (Linux), Podman, or Apple Containers (macOS 26+) installed. Kubernetes support is on the roadmap. ### How is this different from CrewAI or LangGraph? CrewAI and LangGraph are Python libraries — you build your own runtime around them. Crewship is a complete self-hosted platform: one binary, a UI, container isolation, audit trail, budgets, approvals, and credential vault. They're complementary — you can wrap either library inside a Crewship crew. ### Can I run agents without depending on any single cloud LLM? Yes. Crewship orchestrates official CLIs from Anthropic (Claude Code), OpenAI (Codex), Google (Gemini), and OpenCode (open-source). Mix any combination per agent — if one vendor changes terms or has an outage, swap that agent's CLI without touching the rest of the crew. Native Ollama-as-runtime support is on the roadmap. ### Does this work with my Claude Code subscription? Yes. Crewship invokes the claude CLI inside each agent container, authenticated with whatever credentials you mount. If you have a Claude Max / Pro / API plan, Crewship uses the same limits your account has. ### How do I keep LLM costs under control? Paymaster enforces budgets at every level: workspace → crew → mission → agent. Hard stops before the call is made, not after the bill arrives. You can also gate high-cost actions via Harbormaster (human approval). ### What's the upgrade path to Enterprise features (SSO, federation)? The Apache-2.0 core includes everything listed on this page — Journal, Paymaster, Keeper, Harbormaster, Cartographer, Crow's Nest. Enterprise adds SSO/SAML, cross-workspace federation, priority support and a curated marketplace. EE is a drop-in extension, not a fork. ### What happens to my data if I stop using Crewship? Your database is yours — PostgreSQL or SQLite, depending on your setup. Everything lives in `journal_entries` (append-only) and standard relational tables. Export, migrate, archive — no lock-in. ### Why containers? Why not just run agents as processes? Containers are the correct security boundary for untrusted code — this is a pattern from core banking systems. An AI agent executes arbitrary tool calls, writes files, and makes network requests based on LLM output. When prompt injection happens (and it will), the container stops the damage at the boundary: non-root UID, capability-dropped, read-only rootfs optional, network policy configurable. Processes share everything with the host. That is not acceptable for agents. ## Pricing Apache-2.0 core is free forever. Journal, Paymaster, Keeper, Harbormaster, Cartographer, Crow's Nest — all of it, forever. Enterprise (SSO, federation, marketplace, priority support) coming soon. ## Links - Homepage: https://crewship.ai - GitHub: https://github.com/crewship-ai/crewship - Documentation: https://docs.crewship.ai - Discussions & Roadmap: https://github.com/crewship-ai/crewship/discussions - Discord: https://discord.gg/crewship