How langbot-app/langbot works
LangBot is a platform for configuring and running AI-powered bots across instant-messaging services. The primary workflow is an incoming platform event, normalized by an adapter, processed through a configured pipeline and its AI/agent capabilities, then answered through the platform. The management API and web panel configure pipelines, integrations, and workspaces; monitoring surfaces runtime activity. The README documents broad provider, knowledge-base, plugin, and MCP capabilities. Some runtime wiring and those subsystems’ implementations were not sampled, so uncertain links are omitted.
Main entry points: pipelinemgr.py Pipeline runtime · controller HTTP management API · adapters Platform adapters · pipeline.py Pipeline records
Components 16
Chat channels
- Platform adapters
src/langbot/pkg/platform/adapters
Conversation runtime
- Pipeline runtime
src/langbot/pkg/pipeline/pipelinemgr.py - Pipeline stages
src/langbot/pkg/pipeline - Agent runner
src/langbot/pkg/agent/runner - AI and RAG integrations
src/langbot/pkg/api/http/controller/groups/knowledge
Management and extensions
- Plugins and components
src/langbot/pkg/discover/engine.py - MCP resources
src/langbot/pkg/api/mcp - HTTP management API
src/langbot/pkg/api/http/controller - Web management panel
web/src/app/home/pipelines/components/pipeline-form/PipelineFormComponent.tsx - Workspace services
src/langbot/pkg/api/http/service/space.py - Monitoring panel
web/src/app/home/pipelines/components/monitoring-tab/PipelineMonitoringTab.tsx
Persistence and monitoring
- Application database
src/langbot/pkg/persistence/database.py - Pipeline records
src/langbot/pkg/entity/persistence/pipeline.py - Monitoring
src/langbot/pkg/pipeline/monitoring_helper.py
Other components
- Chat user
- IM platforms
Connections 16
- Chat user → IM platforms: sends message
- IM platforms → Platform adapters: delivers event
- Platform adapters → Pipeline runtime: dispatches query
- Pipeline runtime → Pipeline stages: runs stages
- Pipeline runtime → Agent runner: invokes runner
- Pipeline runtime → AI and RAG integrations: uses AI services
- Pipeline runtime → Plugins and components: dispatches extensions
- Pipeline runtime → MCP resources: uses MCP
- Pipeline runtime → Platform adapters: sends response
- Web management panel → HTTP management API: calls management API
- HTTP management API → Workspace services: uses workspace services
- HTTP management API → Pipeline records: manages pipelines
- Pipeline records → Application database: persists records
- Pipeline runtime → Monitoring: records activity
- Monitoring panel → HTTP management API: loads monitoring
- Monitoring → Application database: stores metrics