How singnet/omega works
Omega is documented as a stateful neural-symbolic agent, with this tree exposing its Python integration layers rather than the MeTTa execution loop. The principal user workflow is a message arriving through a configured channel, being handled by the agent runtime, and receiving a response using a selected LLM provider. Supporting branches include authentication, retrieval-backed knowledge, memory portability, and OpenClaw integration. Edges reflect the shown channel/provider registries, imports, and explicit README workflow; the central agent loop and some subsystem wiring are not present in the sampled files.
Main entry points: channels.py Channel registry · lib_llm_ext.py LLM client layer · providers.py Provider registry · memory_export.py Memory portability
Components 28
Communication
- Channel registry
src/channels.py - Channel authentication
channels/auth.py - IRC channel
channels/irc.py - Telegram channel
channels/telegram.py - Slack channel
channels/slack.py - Mattermost channel
channels/mattermost.py - WebSocket chat
channels/wschat.py - Delivery queue
channels/delivery_queue.py
Model and knowledge
- Omega agent loop
- Provider registry
src/providers.py - LLM client layer
providers/lib_llm_ext.py - OpenAI provider
providers/openai.py - OpenRouter provider
providers/openrouter.py - ASI:One provider
providers/asione.py - Custom OpenAI API
providers/openaiapi.py - Knowledge retrieval
src/rag.py - Vector store
- Embedding selection
src/embedding_models.py
Memory and policy
- Memory portability
src/memory_export.py - Memory files
src/fileio.py - Profile policy
profile/policy.py
Extensions and operations
- Configuration
src/config.py - Plugin framework
src/plugin.py - OpenClaw integration
plugins/openclaw/openclaw.py - Web search
src/websearch.py - Agent utilities
src/helper.py
Other components
- Agent user
- LLM services
Connections 41
- Agent user → IRC channel: messages
README.md - Agent user → Telegram channel: messages
README.md - IRC channel → Channel registry: registers channel
channels/irc.py - Telegram channel → Channel registry: registers channel
channels/telegram.py - Slack channel → Channel registry: registers channel
channels/slack.py - Mattermost channel → Channel registry: registers channel
channels/mattermost.py - WebSocket chat → Channel registry: implements interface
channels/wschat.py - Channel registry → Channel authentication: checks identity
src/channels.py - IRC channel → Channel authentication: checks authorization
channels/irc.py - Telegram channel → Channel authentication: checks authorization
channels/telegram.py - Slack channel → Channel authentication: checks authorization
channels/slack.py - Mattermost channel → Channel authentication: checks authorization
channels/mattermost.py - IRC channel → Delivery queue: uses queue
channels/irc.py - Telegram channel → Delivery queue: uses queue
channels/telegram.py - Slack channel → Delivery queue: uses queue
channels/slack.py - Mattermost channel → Delivery queue: uses queue
channels/mattermost.py - Channel registry → Omega agent loop: receives messages
- Omega agent loop → Channel registry: sends replies
README.md - Omega agent loop → Provider registry: requests completion
README.md - Provider registry → OpenAI provider: registers provider
providers/openai.py - Provider registry → OpenRouter provider: registers provider
providers/openrouter.py - Provider registry → ASI:One provider: registers provider
providers/asione.py - Provider registry → Custom OpenAI API: registers provider
providers/openaiapi.py - OpenAI provider → LLM client layer: delegates chat
providers/openai.py - OpenRouter provider → LLM client layer: delegates chat
providers/openrouter.py - ASI:One provider → LLM client layer: uses client layer
providers/asione.py - Custom OpenAI API → LLM client layer: uses client layer
providers/openaiapi.py - LLM client layer → LLM services: calls API
providers/lib_llm_ext.py - Omega agent loop → Knowledge retrieval: queries knowledge
- Knowledge retrieval → Vector store: stores and queries
src/rag.py - Knowledge retrieval → Embedding selection: selects model
src/rag.py - Channel registry → Memory portability: routes export command
src/channels.py - Memory portability → Memory files: reads and writes
src/memory_export.py - Omega agent loop → Profile policy: applies policy
- Omega agent loop → OpenClaw integration: invokes integration
- Omega agent loop → Web search: uses search
- OpenClaw integration → Configuration: reads settings
plugins/openclaw/openclaw.py - Knowledge retrieval → LLM client layer: uses embedding helpers
src/rag.py - Memory portability → Configuration: reads settings
src/memory_export.py - Memory portability → Embedding selection: uses model settings
src/memory_export.py - Memory portability → Agent utilities: uses utilities
src/memory_export.py