Connectors

Connectors are the MCP layer that brings external systems into MindMux.

What connectors do

Connectors let the agent call tools exposed by external services, such as:

• Reading GitHub issues or PR status
• Querying Notion pages
• Creating or updating Linear issues
• Accessing GitLab repositories

They turn external capabilities into structured tools inside conversations, rather than making every integration a separate product surface.

Currently supported connectors

Official connectors

• GitHub: hosted HTTP with bearer token.
• Linear: hosted HTTP with bearer token.
• Notion: OAuth-first connector.
• GitLab: OAuth-first connector, supports self-managed instances.

Custom MCP V1

Besides official connectors, you can manually add custom MCP servers:

• Transport: stdio, legacy SSE (via a local stdio bridge), streamable HTTP.
• Auth: none, bearer, api key header, custom headers.

Management entry points

• Settings > Connectors: manually add, enable, disable, test, and authorize connectors.
• Configuration expert: a conversational entry point that shares the same connector state machine with the Connectors UI.

When connectors take effect

Connector configuration mainly affects newly created sessions. Existing sessions do not hot-reload the MCP tool list.

Boundary with Skills

• Connectors change which external systems MindMux can talk to.
• Skills change how MindMux behaves (what the AI does in specific situations).

Keeping them separate makes both the architecture and the settings UI clearer.

Settings entry points

The Settings sidebar is a flat list of 6 items:

1. General
2. Runtime (provider profiles)
3. Experts
4. Connectors
5. Skills
6. Workspace