How thiagofinch/mega-brain works
Mega Brain turns user-provided material into organized, searchable knowledge, then supports evidence-grounded strategic decisions. The documented flow is ingest, process and index, query, and deliberate; separate capabilities include video analysis and JARVIS voice interaction. The graph preserves major README features and indexed core modules. Some implementation details, especially Conclave and unsampled pipeline stages, are documented but not shown in source excerpts, so those links are identified as documentation-based or left unverified.
Main entry points: batch_auto_creator.py Batch Processing · orchestrator.py Voice Orchestrator · pipeline.py RAG Pipeline · ingest-with-entity-discovery.py Ingestion Entry
Components 19
Ingestion and Processing
- Ingest Command
- Ingestion Entry
scripts/ingest-with-entity-discovery.py - Video Analysis
engine/intelligence/pipeline/video/pipeline.py - Batch Processing
engine/intelligence/pipeline/batch_auto_creator.py - Bucket Routing
engine/intelligence/pipeline/pipeline_router.py - LLM Routing
engine/intelligence/pipeline/mce/llm_router.py
Knowledge and RAG
- Local Knowledge
engine/intelligence/memory_manager.py - RAG Pipeline
engine/intelligence/rag/pipeline.py - Adaptive Retrieval
engine/intelligence/rag/adaptive_router.py - Hybrid Search
engine/intelligence/rag/hybrid_query.py - Hybrid Index
engine/intelligence/rag/hybrid_index.py - Bucket Queries
engine/intelligence/rag/bucket_query_router.py
Decision Support
- Conclave Council
engine/intelligence/pipeline/mce
JARVIS Voice
- Voice Application
engine/jarvis/voice/main.py - Voice Orchestrator
engine/jarvis/voice/orchestrator.py - Speech Recognition
engine/jarvis/voice/stt_handler.py - Brain Connector
engine/jarvis/voice/mega_brain_connector.py - Speech Output
engine/jarvis/voice/tts_handler.py
Other components
- User
Connections 20
- User → Ingest Command: submits material
- Ingest Command → Ingestion Entry: starts ingestion
- Ingestion Entry → Video Analysis: uses video pipeline
scripts/ingest-with-entity-discovery.py - Ingestion Entry → Batch Processing: creates batches
scripts/ingest-with-entity-discovery.py - Batch Processing → Bucket Routing: routes content
- Batch Processing → LLM Routing: extracts insights
- Batch Processing → Local Knowledge: enriches memory
engine/intelligence/pipeline/batch_auto_creator.py - User → RAG Pipeline: asks questions
- RAG Pipeline → Adaptive Retrieval: routes retrieval
engine/intelligence/rag/pipeline.py - Adaptive Retrieval → Hybrid Search: runs hybrid search
engine/intelligence/rag/adaptive_router.py - Adaptive Retrieval → Hybrid Index: uses index
engine/intelligence/rag/adaptive_router.py - Bucket Queries → Hybrid Search: searches buckets
engine/intelligence/rag/bucket_query_router.py - Bucket Queries → Hybrid Index: selects bucket index
engine/intelligence/rag/bucket_query_router.py - RAG Pipeline → Local Knowledge: builds memory context
engine/intelligence/rag/pipeline.py - User → Conclave Council: requests deliberation
- Conclave Council → RAG Pipeline: grounds decisions
- Voice Application → Voice Orchestrator: starts
engine/jarvis/voice/main.py - Voice Orchestrator → Speech Recognition: receives transcripts
engine/jarvis/voice/orchestrator.py - Voice Orchestrator → Brain Connector: uses brain state
engine/jarvis/voice/orchestrator.py - Voice Orchestrator → Speech Output: speaks responses
engine/jarvis/voice/orchestrator.py