Skip to content

webfetch v0.1.0

Choose a tag to compare

View all tags
@firish firish released this 15 Jul 02:24
· 19 commits to main since this release
v0.1.0
d1ce233

Initial release. Web search for LLM agents that you run yourself.

Benchmarked against hosted web_search tools and the popular search APIs (same agent loop, same questions, same judge): matches hosted accuracy at roughly a third of the cost, with 5-8x fewer input tokens per search. Full tables and methodology in the README.

Highlights:

  • web_search tool for agent loops with a crash-proof handler
  • Multi-engine fusion (DDG/Brave/Serper/Tavily), hybrid semantic ranking, sentence-level compression (half the tokens, zero measured recall loss)
  • Semantic paraphrase cache with volatility-aware TTLs - repeat queries cost nothing
  • Circuit breakers and failover for engines that silently block
  • Cost receipts: webfetch-savings shows what you did not pay
  • MCP server: claude mcp add webfetch webfetch-mcp

Install: pip install "webfetch-llm[all]" (pip >= 24; PyPI name is webfetch-llm, import name is webfetch)

See CHANGELOG.md for the full list.

Assets 2
Loading