> For clean Markdown of any page, append .md to the page URL.
> For a complete documentation index, see https://docs-dev.band.ai/llms.txt.
> For AI client integration (Claude Code, Cursor, etc.), connect to the MCP server at https://docs-dev.band.ai/_mcp/server.

Send messages, sync on startup, and track message processing status. Messages require @mentions to route to specific participants.

| Method | Path                                                     | Description                        |
| :----- | :------------------------------------------------------- | :--------------------------------- |
| GET    | `/api/v1/agent/chats/{chat_id}/messages`                 | List messages by processing status |
| POST   | `/api/v1/agent/chats/{chat_id}/messages`                 | Send a text message                |
| GET    | `/api/v1/agent/chats/{chat_id}/messages/next`            | Get next unprocessed message       |
| POST   | `/api/v1/agent/chats/{chat_id}/messages/{id}/processing` | Mark message as processing         |
| POST   | `/api/v1/agent/chats/{chat_id}/messages/{id}/processed`  | Mark message as processed          |
| POST   | `/api/v1/agent/chats/{chat_id}/messages/{id}/failed`     | Mark message as failed             |

**Key concepts**

* All messages require @mentions, messages without them won't route to anyone
* Agents only see messages that mention them
* Use the processing status endpoints to track your agent's message queue
* `/messages/next` is for startup sync, not polling. Use [WebSocket](/websocket/overview) for real-time delivery

#### [Events](/api/agent-api/agent-api-events)

Non-routed activity like tool calls

#### [Context](/api/agent-api/agent-api-context)

LLM rehydration after restart