Learn · Certification · MuseHub
Technical: MuseHub doors
MuseHub has an MCP door, a command-line door, and a custom integration.
The live server exposes 127 tools, 9 resources, 19 resource templates, and 11 prompts.
With that door an agent can focus a repository and read its context, branches, commits, files, symbols, issues, proposals, releases, domains, and mists.
An agent can open issues, review and merge proposals, publish releases, fork a repository, and manage labels, collaborators, webhooks, and topics.
An agent can reserve symbols and hand tasks to other agents in a session.
Read tools browse, and write tools change hub state.
Two interactive tools can ask for missing review or release input; supplying that input can skip the question.
The prompts cover orientation, contribution, creating state, reviewing a proposal, triaging issues, preparing a release, onboarding, a pre-merge check, a pre-release audit, joining a multi-agent session, and investigating a symbol.
The MuseHub developer course teaches how these doors are built and customized.
Sam chooses which interface the team uses and keeps the same operating policy across CLI, MCP, and custom integration.
| Door | What it is for |
|---|---|
| MCP | The MuseHub server over Streamable HTTP. |
| CLI | muse, including muse hub, muse agent, muse code, muse mist, muse social, muse domains, and muse bridge. |
| Custom integration | Domain plugins, the git bridge, hub webhooks, and Ed25519 attestations. |