One Linux container per crew.
Agents in a crew share a working environment. Separate crews can use different images, tools, and credentials. The container is the runtime boundary; an agent identity is not a separate container.
Explore devcontainersA real workspace. On your hardware.
Run Crewship where your team works. Each crew gets an isolated Linux container with its own tools, dependencies, and assigned access.
Use the tools the work requires.
Configure a Debian, Ubuntu, or glibc-based Python image. Add CLIs, database clients, and development tools. Connect external services through their APIs or configured MCP tools.
Configure an environmentKeep the workspace with your company.
Host the application, working files, memory, and journal on infrastructure you control. You remain responsible for host access, backups, updates, and the credentials you assign.
Plan your deploymentTwo crews. Two working environments.
- Operations · team documents, spreadsheet tools, assigned service access
- Engineering · repository, Python, PostgreSQL client
- Shared platform · conversations, issues, routines, and a journal
Self-hosting does not make every connection local. Cloud AI and connected services receive the data your configured tools send.
One platform.
See the real workspace.
Actual screenshots from our development instance, with demo data.
Give every crew a purpose.
Sales, finance, and operations crews in the actual Crewship workspace.
Captured from dev1 · Demo workspace · September 2026START WITH ONE CREW
Get started