yams

You can index code and documents to enable fast semantic and full-text search by initializing a repository and building a persistent memory index. This process extracts metadata and maps relationships for contextual recall across large repositories.

Persistent memory for LLMs uses content-addressed storage and deduplication to retain project context. It supports agent workflows through snapshot management and MCP integration, enabling fast retrieval of contextual relationships across codebases.

Yes, you can perform both full-text and vector search over project memory. The system indexes code and documents to support fast semantic search alongside traditional text matching for comprehensive contextual recall.

YAMS is designed for software development teams needing fast retrieval of context and relationships across large codebases. It provides end-to-end indexing, metadata support, and knowledge graph visualization for large repositories.

Knowledge graph extraction is performed by indexing your code and documents to map relationships and metadata. This builds a persistent memory index that visualizes connections and enables contextual recall across the project.

The modular extension architecture supports optional scripts, references, and assets during code indexing. Limitations depend on the scale of your repositories and the specific metadata structures required for accurate knowledge graph visualization.