Wigolo MCP Server: Setup for Claude, Cursor and Other Apps

• Search the web and get cited evidence

  Searches many engines at once, ranks the results on your computer, and returns scored excerpts with citation IDs. You can pass several queries together and limit by site or date.

  Try: “Search the Rust docs and blog for how async functions in traits work.”

  Network · search

• Read one page as clean markdown

  Loads a web page as markdown, moving up to a real browser when the page needs JavaScript or blocks plain requests. It can also click and type on the page before reading it.

  Try: “Fetch the Data Cache section of the Next.js caching docs.”

  Write · fetch

• Crawl many pages into the local cache

  Follows links or a sitemap from a start page, respecting robots.txt, and saves the pages in the local cache. A map mode lists the URLs without their content.

  Try: “Crawl the Astro docs guides from the sitemap, up to 40 pages.”

  Files · crawl

• Pull tables and structured data from a page

  Pulls tables, metadata, JSON-LD, brand details or fields matching a schema you give out of a page. Most modes need no LLM.

  Try: “Extract every table from the PostgreSQL versioning page.”

  Network · extract

• Search, check or clear what it already saw

  Searches pages wigolo already saved, shows cache totals, checks saved pages for changes, and can delete matching entries.

  Try: “How many pages have you cached, and are any of the Node.js release pages changed?”

  Delete · cache

• Find pages similar to a URL or topic

  Finds pages like a URL or a described topic by combining the local cache with live web search. It works best after a crawl.

  Try: “Find pages similar to 12factor.net/config.”

  Network · find_similar

• Answer a question from many sources

  Splits a question into sub-queries, searches and cross-checks the sources, and returns a structured brief with per-claim sources and gaps. With no LLM set, it returns the brief for your assistant to write up.

  Try: “How do Deno and Bun differ in Node.js compatibility?”

  Network · research

• Gather data on its own within a time budget

  Plans searches and URLs from your prompt, fetches pages in parallel, optionally extracts fields you define, and logs every step. It stops at a page and time limit.

  Try: “Gather the current LTS support windows for Node.js releases as JSON.”

  Network · agent

• Compare two versions of a page or text

  Compares a live page with its cached copy, two URLs, or two pieces of markdown, and returns a patch, a section list or a count of changes.

  Try: “What changed on the Node.js previous-releases page since last time?”

  Network · diff

• Watch pages for changes

  Registers jobs that re-check pages on an interval and report changes inline or to a webhook address. Jobs only run while a wigolo session or server is active.

  Try: “Watch the Go release page every hour and tell me when it changes.”

  Delete · watch

Watch out for

• The browser can click and type on pages

  The fetch tool accepts click and type steps, and the source marks it as able to change things on the page. Do not point it at sites where you are signed in unless you mean to. github.com

• It can use your saved browser login

  With the use_auth option, fetch and crawl use your stored browser session. Leave it off unless you need a logged-in page. github.com

• It can delete its own cache and watch jobs

  The cache tool can delete saved pages that match a filter, and the watch tool can delete its own jobs. Both only touch wigolo's local store. github.com

• Watch jobs can call a webhook

  The watch tool can send change reports to a URL you provide. Private and loopback addresses are refused. github.com

• Its HTTP server needs a token off localhost

  The optional serve mode is open only on 127.0.0.1. Bound to any other address it refuses to start without a token. github.com

Security advisories