ref-tools/ref-tools-mcp

Helping coding agents never make mistakes working with public or private libraries without wasting the context window.

45
/ 100
Emerging

Implements an MCP server with two tools—semantic search and URL fetching—that intelligently filters documentation results based on session search history to return only the most contextually relevant sections, typically 5k tokens maximum. Uses adaptive multi-step search patterns where agents refine queries based on intermediate results, avoiding the token waste of traditional pagination and over-fetching. Designed for integration with Claude Code, Cursor, and OpenAI deep research, with both HTTP (recommended) and stdio transport options.

1,007 stars.

No Package No Dependents
Maintenance 10 / 25
Adoption 10 / 25
Maturity 9 / 25
Community 16 / 25

How are scores calculated?

Stars

1,007

Forks

65

Language

TypeScript

License

MIT

Last pushed

Jan 09, 2026

Commits (30d)

0

Get this data via API

curl "https://pt-edge.onrender.com/api/v1/quality/mcp/ref-tools/ref-tools-mcp"

Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.