How insightos-community/semantic-framework works
InsightOS Semantic is a framework for embodied AI and robotic applications. Its principal workflow is a user request entering the Semantic Server, where an agent interprets and plans the task, workflow orchestration coordinates execution, and robot services dispatch work to Pilots and robot runtimes. Execution feedback and operational records support monitoring and continued capability development. The README documents major capabilities including collaboration, semantic world models, simulation, and skill evolution. Some internal relationships are documented only at a high level; the graph avoids asserting unsampled call paths.
Main entry points: runtime/service.go Agent runtime · http HTTP and WS API · robot/service.go Robot service · scheduler.go Workflow scheduler
Components 18
Application interaction
- HTTP and WS API
internal/server/http - Authentication
internal/server/auth
Task orchestration
- Agent runtime
internal/agent/runtime/service.go - Agent kernel
internal/agent/kernel - Workflow scheduler
internal/workflow/scheduler.go - Human interaction
internal/interaction/service.go
Robot execution
- Robot service
internal/robot/service.go - Semantic Pilot
cmd/semantic-pilot/main.go - Robot runtime
internal/robotruntime/orchestrator.go - Robot Skills
internal/skill/store.go
World and simulation
- Semantic world model
internal/tool/builtin/map_query.go - Simulation services
internal/simulation/service.go
Shared platform
- MCP integrations
internal/mcpregistry/registry.go - Persistent state
internal/store
Other components
- Studio user
- Semantic Studio
- Model providers
- Robots and devices
Connections 19
- Studio user → Semantic Studio: describes goal
README.md - Semantic Studio → HTTP and WS API: sends requests
README.md - HTTP and WS API → Authentication: routes auth
internal/server/http/router.go - HTTP and WS API → Agent runtime: routes requests
- Agent runtime → Agent kernel: runs agent
internal/agent/runtime/service.go - Agent runtime → Model providers: requests reasoning
README.md - Agent runtime → Workflow scheduler: organizes tasks
README.md - Agent runtime → Human interaction: requests approval
- Workflow scheduler → Robot service: coordinates robot work
README.md - Robot service → Semantic Pilot: dispatches actions
README.md - Semantic Pilot → Robot runtime: runs capabilities
README.md - Robot runtime → Robots and devices: connects devices
README.md - Robot service → Robot Skills: manages skills
internal/robot/service.go - Agent runtime → Semantic world model: queries semantics
- HTTP and WS API → Simulation services: routes simulation
internal/server/http/router.go - Agent runtime → MCP integrations: uses tools
internal/agent/runtime/service.go - Agent runtime → Persistent state: persists runtime state
internal/agent/runtime/service.go - Workflow scheduler → Persistent state: reads and writes
internal/workflow/scheduler.go - Robot service → Persistent state: records execution
internal/robot/service.go