The bridge now supports routing to any agent in the AI Maestro mesh network using the @AIM:agent-name convention.
@AIM:backend-api check the server status
@AIM:graph-query find all API endpoints
Messages without @AIM: prefix route to the default agent (configurable via DEFAULT_AGENT env var).
Responses are prefixed with the agent name:
🤖 [backend-api] Server is healthy, all endpoints responding.
Currently the bridge connects to a single Slack workspace. Each workspace requires its own Bot Token (xoxb-...).
Option A: Multiple Bridge Instances
- Run separate bridge processes per workspace
- Each with its own
.envfile (.env.workspace1,.env.workspace2, etc.) - Simple but requires managing multiple processes
Option B: Single Bridge, Multiple Tokens
- Modify bridge to accept array of workspace configs
- Single process handles all workspaces
- More complex but cleaner operationally
- Go to api.slack.com/apps
- Settings → Manage Distribution → Enable distribution
- Install to Another Workspace
- Get new Bot Token (
xoxb-...) - Configure and run bridge for that workspace
- Agent routing via @AIM:agent-name ✅
- Multi-workspace support (see above)
- PM2 or systemd service for persistent running
- Better error handling for rate limits (429)
- Retry logic with exponential backoff
- Message queue to prevent rate limiting
- Slack thread context preservation across sessions
- File/image attachment support
- Slash commands (
/lola help)
- Web dashboard for bridge status
- Message history/logging
- Per-channel or per-user preferences
- Scheduled messages
- 429 rate limiting - need to investigate source and add throttling