How googol23/bd1 works
BD-1 is a local text-chat and simulated robot-control prototype, with a separate local visual interface and an offline-capable voice-session boundary. Text requests route to chat, metadata, capability, clarification, or simulated-action handling; actions pass deterministic policy checks before changing in-memory simulation state. Receipts are audited locally. The graph also preserves dialogue, intent, and voice components, but their unshown internal wiring is omitted. The README documents no hardware or persistent state; unsampled modules and integrations are not inferred beyond indexed references.
Main entry points: core.py Text request handler · server.py Visual API server · policy.py Action policy · llm.py Ollama adapter
Components 19
User interfaces
- Text chat entry
app.py - Visual API server
bd1/ui/server.py - Visual UI assets
bd1/ui/index.html - Voice session
bd1/audio/session.py - Mock audio adapters
bd1/audio/mock.py
Conversation handling
- Text request handler
core.py - Request router
bd1/router.py - Intent handling
bd1/intent.py - Dialogue context
bd1/dialogue.py - Model metadata
bd1/metadata.py
Simulation and policy
- Action policy
bd1/policy.py - Tool registry
bd1/tools/registry.py - Simulation controller
bd1/controller.py - Simulation state
bd1/state.py
Local services
- Ollama adapter
bd1/llm.py - Audit log
bd1/audit.py
Other components
- User
- Local Ollama
- Local browser
Connections 21
- User → Text chat entry: sends text
README.txt - Text chat entry → Text request handler: calls
app.py - Text request handler → Request router: routes via
core.py - Text request handler → Dialogue context: builds context
core.py - Text request handler → Model metadata: fetches metadata
core.py - Text request handler → Action policy: dispatches actions
core.py - Text request handler → Ollama adapter: requests chat
core.py - Ollama adapter → Local Ollama: calls local API
bd1/llm.py - Text request handler → Audit log: records result
core.py - Voice session → Text request handler: handles confirmed text
bd1/audio/session.py - Voice session → Mock audio adapters: uses injected adapters
bd1/audio/mock.py - Action policy → Tool registry: validates tools
bd1/policy.py - Action policy → Simulation controller: executes action
bd1/policy.py - Simulation controller → Simulation state: reads and updates
bd1/controller.py - User → Visual API server: uses visual view
README.txt - Visual API server → Visual UI assets: serves assets
bd1/ui/server.py - Visual API server → Action policy: dispatches actions
bd1/ui/server.py - Visual API server → Audit log: records result
bd1/ui/server.py - Local browser → Visual API server: requests local API
- Intent handling → Ollama adapter: references
bd1/intent.py - Intent handling → Tool registry: references
bd1/intent.py