Webfetch (firish) MCP Server

Web search for LLM agents that you run yourself - up to 8x fewer input tokens and 3x lower cost than hosted web_search, at the same accuracy. It runs locally over stdio via the published package.

People connecting search tools to Claude, Cursor, VS Code, or another MCP client. The project is written in Python.

VERIFIED ACTIVE

LAST COMMIT 2026-07-23 · ★ 54 · #16 OF 47 MAINTAINED SEARCH · VERIFIED 2026-08-25

MIT · Python servers · how we verify → /methodology

01 · Install Webfetch (firish)

Claude Code

claude mcp add firish-webfetch -- uvx webfetch-llm

Claude Desktop / Cursor / VS Code - add to config

{
  "mcpServers": {
    "firish-webfetch": {
      "command": "uvx",
      "args": [
        "webfetch-llm"
      ]
    }
  }
}

Same JSON for Cursor. For VS Code, rename the top-level key from `mcpServers` to `servers`.

Using another client? Same JSON, different key

Claude Desktop · mcpServers

Cursor · mcpServers

VS Code · servers

Windsurf · mcpServers

Zed · context_servers

Cline · mcpServers

Roo Code · mcpServers

Continue · mcpServers

LibreChat · mcpServers

Gemini CLI · mcpServers

Codex CLI · mcp_servers

Full setup guides: every client.

02 · Evidence

Security posture

What to check before giving this server access to your agent - from the registry, GitHub, and our own probes. We don't score safety; we show what's verifiable.

runs as local process (stdio) - runs on your machine with your user's permissions

license MIT - declared in the repository

pypi package webfetch-llm - check the name against the project README before installing (PyPI has no namespace ownership)

registry namespace io.github.firish is GitHub-verified and matches the repo owner

03 · What Webfetch (firish) can do

Prose above is summarized from the project's README and registry record - no invented capabilities.

Limitations (from the README)

- Latency. A fresh search takes 10-40 seconds - real pages get fetched and ranked locally. Hosted search returns in ~10s, snippet APIs in ~6s. Cache hits are instant. If you need sub-second search and do not care about cost, this is not your tool. - The cache is single-process. One agent loop, one MCP server, or one notebook at a time; a shared team cache backend is on the roadmap. - Engine free tiers are real quotas

Latest releases

v0.1.3 · 2026-07-19

Distribution release: webfetch is now listed in the official MCP Registry and installable as a Claude Code plugin. · Official MCP Registry listing as io.github.firish/webfetch, auto-published by CI on every release…

v0.1.2 · 2026-07-19

The field-tested release: everything in it came from real usage findings. · New tools and levers: · fetch_url - one cited page in full, under a budget, instant for pages the pipeline has already fetched; public-URL…

v0.1.0 · 2026-07-15

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…

04 · Who maintains Webfetch (firish)

webfetch is maintained by firish. It's the only MCP server we track from this author; the repo dates to Apr 2026.

05 · Facts

category
search - ranked #16 of 47 actively-maintained search servers as of 2026-08-25.
release cadence
3 releases in the last 90 days (latest 2026-07-19)
registry
io.github.firish/webfetch (active, first published 2026-07-19)
packages
pypi:webfetch-llm

06 · Webfetch (firish) FAQ

What is Webfetch (firish)?

Web search for LLM agents that you run yourself - up to 8x fewer input tokens and 3x lower cost than hosted web_search, at the same accuracy. It runs locally over stdio via the published package.

Is Webfetch (firish) still maintained?

Yes - as of 2026-08-25, its last commit was 2026-07-23 and it shipped 3 releases in the last 90 days. We re-verify nightly.

How do I install Webfetch (firish)?

Run `uvx webfetch-llm`. You can also paste the ready-made client config above.

Does Webfetch (firish) run locally?

Yes - it's a stdio server: it runs on your machine (via uvx) with your user's permissions. Your data stays local unless the server itself calls external APIs.

07 · Alternatives to Webfetch (firish)