ontorata/ratary

Overview

A TypeScript MCP server that gives AI assistants persistent coding memory. It uses hybrid search and a knowledge graph to store and retrieve context, aiming for token-efficient recall. Runs over stdio.

Best for

Best for
Developers experimenting with persistent memory for MCP-based AI assistants

Use cases

• Store project context for an AI coding assistant
• Retrieve relevant code knowledge via hybrid search
• Maintain a knowledge graph of coding concepts across sessions

Notes

A TypeScript MCP server that gives AI assistants persistent coding memory. It uses hybrid search and a knowledge graph to store and retrieve context, aiming for token-efficient recall. Runs over stdio.

1 stars on GitHub. Last updated 2026-08-10. Licensed MIT.

Use cases

• Store project context for an AI coding assistant
• Retrieve relevant code knowledge via hybrid search
• Maintain a knowledge graph of coding concepts across sessions

Pros

• Token-efficient context retrieval
• Combines hybrid search with knowledge graph
• Simple stdio-based MCP integration

Cons

• Very early stage with only 1 star
• Limited adoption and community
• No evidence of production readiness

Indexed from awesome-mcp-servers-punkpeye and enriched against its public facts.