Matrix with matrix-xmsg
matrix-xmsg connects Matrix support rooms to a dedicated
expert agent session behind xmsg. It registers with the bus as the matrix-xmsg service and
uses xmsg messages and replies to carry the conversation.
flowchart LR M["Matrix room"] --> B["matrix-xmsg: admission and context"] B --> X["xmsg service connection"] X --> A["Expert agent session"] A -->|"reply"| X X --> B B -->|"Matrix thread response"| M
A bridge, with its own admission policy
Section titled “A bridge, with its own admission policy”The bridge handles room and sender admission, mention and thread triggers, size/rate limits, context assembly and response placement. It can operate in trusted or public admission modes. Public-room content carries sender and tier provenance; a Matrix user’s text does not become an authorized local instruction merely because the bridge delivered it.
The attested bus sender identifies the bridge service. The bridge’s envelope identifies the Matrix participants and context inside that message. Configure the bridge’s outside-facing security and the expert’s permissions deliberately; xmsg does not authenticate individual Matrix accounts.
For supported triggers, admission modes, guard integration, encryption and deployment, use the Matrix bridge README and its configuration sources. See also provenance versus authority.
Palettes adapted from Catppuccin (Macchiato) (MIT), Tokyo Night (Apache-2.0), gruvbox (MIT), Catppuccin (Latte) (MIT), Rosé Pine (Dawn) (MIT).