SCIP-based code intelligence pipeline for the Archon RAG system. Provides sync services, vector storage, code graph indexing, and an MCP server across platform repositories.
The knowledge base deploys as a set of services in a dedicated namespace:
- PostgreSQL — persistent storage for document state and code graph
- Qdrant — vector store for embedding-based retrieval
- Embedding service — generates vector embeddings from source documents
- Graph service — GraphQL API over SCIP-derived code relationships
- Query service — unified query interface across vector and graph stores
- MCP server — Model Context Protocol endpoint for agent access
.kiro/docs/- Complete documentation skeleton optimized for RAG retrievalCLAUDE.md- Documentation contract and standards- All required documentation files (overview, architecture, operations, api, data-models, faq)
.kiro/steering/archon-docs.md- Always-active steering that enforces documentation standards- Automatic documentation maintenance across all Kiro tasks
- RAG-friendly structure and provenance tracking
- Use this template to create a new repository
- Customize the documentation files under
.kiro/docs/for your project - Work with Kiro - the steering will automatically maintain documentation standards
All documentation follows these principles:
- Grounded in code - Every statement references actual code or infrastructure
- RAG-friendly - Structured for optimal retrieval (400-800 token sections)
- Provenance - Clear "Source" references to relevant files
- No hallucinations - Only documented, verifiable behavior
- Always current - Updated alongside code changes
This repository is configured to be ingested by Archon, which reads all Markdown files under .kiro/docs/ to build mental models for sourcing code and architectural information.
See CLAUDE.md for the complete documentation contract.
[Add your license]