How foreactlabs/openexecutive works
Open Executive is a configurable AI executive assistant: a person sends work through chat or a connected channel; the API/orchestrator routes it to specialist agents and tools, using company knowledge and memory, then returns a response or takes approval-governed action. It also supports proactive briefings, alerts, monitoring, scheduled workflows, browser use and MCP clients. The graph retains README capabilities even where their implementation was not sampled; those branches are linked only where README or excerpts establish the relationship.
Main entry points: orchestrator Executive orchestrator · pipeline.py Approvals and alerts · episodic.py Episodic memory · main.py FastAPI
Components 20
User surfaces
- Connected channels
packages/core/openexecutive/integrations - Web application
packages/ui/src/app - FastAPI
packages/core/openexecutive/api/main.py - MCP server
packages/core/openexecutive/mcp_server/server.py
Executive runtime
- Executive orchestrator
packages/core/openexecutive/orchestrator - Specialist council
packages/core/openexecutive/agents - Prompt assembly
packages/core/openexecutive/prompts/cache_manager.py
Context and memory
- Knowledge retrieval
packages/core/openexecutive/knowledge - Episodic memory
packages/core/openexecutive/memory/episodic.py - Personal memory
packages/core/openexecutive/memory/honcho_client.py - People registry
packages/core/openexecutive/people/store.py
Delegation and governance
- Mail delegation
packages/core/openexecutive/delegation - Approvals and alerts
packages/core/openexecutive/alerts/pipeline.py - Browser actions
packages/core/openexecutive/browser - Scheduled workflows
packages/core/openexecutive/workflows - Workspace services
packages/core/openexecutive/integrations/workspace
Proactive operations
- Briefings and nudges
packages/core/openexecutive/scheduler/nudge_engine.py - Watch monitoring
packages/core/openexecutive/monitoring/pipeline.py - Audit trail
packages/core/openexecutive/audit/context.py
Other components
- Person
Connections 24
- Person → Web application: uses
- Person → Connected channels: messages
- Web application → FastAPI: requests
packages/ui/src/app/api/backend/[...path]/route.ts - FastAPI → Executive orchestrator: routes turns
packages/core/openexecutive/api/main.py - MCP server → Executive orchestrator: runs turns
packages/core/openexecutive/mcp_server/server.py - Connected channels → Executive orchestrator: delivers messages
- Executive orchestrator → Specialist council: consults
packages/core/openexecutive/orchestrator/router.py - Executive orchestrator → Prompt assembly: builds context
- Executive orchestrator → Knowledge retrieval: retrieves context
packages/core/openexecutive/orchestrator/router.py - Executive orchestrator → Episodic memory: records decisions
packages/core/openexecutive/memory/episodic.py - Executive orchestrator → Personal memory: recalls peers
packages/core/openexecutive/memory/honcho_client.py - Prompt assembly → Episodic memory: reads initiatives
packages/core/openexecutive/prompts/cache_manager.py - Executive orchestrator → Mail delegation: delegates tasks
- Executive orchestrator → Approvals and alerts: raises actions
- Approvals and alerts → People registry: routes by roster
packages/core/openexecutive/alerts/pipeline.py - Executive orchestrator → Browser actions: uses browser
- Executive orchestrator → Scheduled workflows: runs workflows
packages/core/openexecutive/mcp_server/server.py - Mail delegation → Workspace services: uses mail services
- Briefings and nudges → Episodic memory: reads commitments
packages/core/openexecutive/scheduler/nudge_engine.py - Briefings and nudges → Approvals and alerts: creates cards
packages/core/openexecutive/scheduler/nudge_engine.py - Watch monitoring → Approvals and alerts: promotes signals
packages/core/openexecutive/monitoring/pipeline.py - Watch monitoring → Episodic memory: reads initiatives
packages/core/openexecutive/monitoring/pipeline.py - FastAPI → MCP server: mounts service
packages/core/openexecutive/api/main.py - Executive orchestrator → Audit trail: logs activity