How garysychan/ai-os works
AI OS Control Plane is a governed framework for coordinating task state, agent roles, policy checks, workflow execution, and auditable runtime evidence. A user enters through the HTTP API or CLI; the API exposes control-plane and workflow operations, while the workflow engine can delegate lifecycle work to the controller, which dispatches requests through the agent runtime. Tasks and workflow definitions provide the domain state; execution adapters and persistence support controlled operations and recovery. The README documents broader governance, scheduling, monitoring, and collaboration capabilities; unsampled links are omitted where source evidence is unavailable.
Main entry points: services.py API Services · workflows/engine.py Workflow Engine · cli.py Control CLI · controller/engine.py Controller Engine
Components 18
User Interfaces
- HTTP API
src/ai_os/api/app.py - API Services
src/ai_os/api/services.py - Control CLI
src/ai_os/cli.py
Task and Governance
- Task Model
src/ai_os/tasks/models.py - Governance Checks
src/ai_os/governance/checker.py - Workflow Definitions
src/ai_os/workflows/definitions.py
Orchestration Runtime
- Workflow Engine
src/ai_os/workflows/engine.py - Controller Engine
src/ai_os/controller/engine.py - Agent Runtime
src/ai_os/agents/runtime.py - Execution Engine
src/ai_os/execution/engine.py
Policies and Integrations
- Adapter Services
src/ai_os/adapters/service.py - Tool Services
src/ai_os/tools/service.py - Scheduler
src/ai_os/scheduler/service.py - Monitoring
src/ai_os/monitoring/service.py
Persistence and Audit
- Observability
src/ai_os/observability/service.py - SQLite Persistence
src/ai_os/persistence/sqlite_store.py - Workflow Store
src/ai_os/workflows/store.py
Other components
- Operator
Connections 22
- Operator → HTTP API: submits requests
src/ai_os/api/app.py - Operator → Control CLI: runs commands
src/ai_os/cli.py - HTTP API → API Services: routes operations
src/ai_os/api/app.py - API Services → Governance Checks: checks consistency
src/ai_os/api/services.py - API Services → Workflow Definitions: validates workflows
src/ai_os/api/services.py - API Services → Workflow Engine: dry-runs workflows
src/ai_os/api/services.py - Workflow Engine → Controller Engine: delegates lifecycle
src/ai_os/workflows/engine.py - Controller Engine → Agent Runtime: dispatches requests
src/ai_os/controller/engine.py - Workflow Engine → Task Model: advances task state
src/ai_os/workflows/engine.py - Controller Engine → Task Model: updates task status
src/ai_os/controller/engine.py - Execution Engine → Adapter Services: resolves adapters
src/ai_os/execution/engine.py - API Services → Execution Engine: validates plans
src/ai_os/api/services.py - API Services → Agent Runtime: lists agents
src/ai_os/api/services.py - API Services → SQLite Persistence: uses repositories
src/ai_os/api/services.py - API Services → Monitoring: reads metrics
src/ai_os/api/services.py - API Services → Observability: queries audit
src/ai_os/api/services.py - Workflow Engine → Workflow Store: stores sessions
src/ai_os/workflows/engine.py - Scheduler → Agent Runtime: uses agent roles
src/ai_os/scheduler/service.py - Observability → SQLite Persistence: stores runtime events
- Control CLI → Scheduler: exposes commands
- Control CLI → Execution Engine: exposes commands
- Tool Services → Adapter Services: uses adapters
src/ai_os/tools/service.py