Refetch
Refetch is an MCP server that lets agents fetch pages as markdown, search the web and news, and extract structured data via refetch.cloud.
About
io.github.lowenbjer/refetch is a hosted MCP server that equips AI agents with production-grade web access: pull pages as markdown, run web and news searches, and extract structured fields without maintaining headless browsers or brittle scrapers. developers in the Idea phase use it to compare competitors, read documentation, summarize announcements, and ground agent answers in current sources while validating what to build next. The Refetch remote endpoint uses streamable HTTP and an API key header, which keeps setup lightweight compared with self-hosted Playwright stacks. It can also support integrations when agents need live data, but the canonical Skillselion placement is research because discovery is the primary solo-developer story. Beginner-friendly onboarding is advertised via a free console key. Refetch is a web retrieval MCP service, not an on-device browsing skill or static dataset.
- Remote streamable-http MCP at https://refetch.cloud/mcp
- Fetch URLs as markdown for LLM-friendly consumption
- Web search and news search tools for agents
- Structured data extraction from pages
- Requires X-API-Key (ftch_ prefix); free key via refetch.cloud console
Refetch by the numbers
- Data as of Jul 7, 2026 (Skillselion catalog sync)
claude mcp add --transport http refetch https://refetch.cloud/mcp --header "X-API-Key: YOUR_X_API_KEY"Add your badge
Show developers this MCP server is listed on Skillselion. Paste this into your README.
| Transport | HTTP |
|---|---|
| Auth | Required |
What it does
Give your agent markdown page fetches, web and news search, and structured extraction via refetch.cloud without building scrapers yourself.
Who is it for?
Best when you're doing competitor scans, doc lookups, and news-aware research inside Claude Code, Cursor, or Codex.
Skip if: Skip if you need fully offline research, heavy authenticated portal scraping, or ship-phase load testing of your own sites.
What you get
After you add your ftch_ API key, your agent can ground research and validation tasks on current markdown pages and search results from Refetch.
- Markdown representations of fetched URLs for agent reasoning
- Web and news search result sets usable in agent prompts
- Structured extracts from target pages per Refetch tool responses
By the numbers
- Manifest version 1.0.0
- Remote endpoint https://refetch.cloud/mcp (streamable-http)
- Required header X-API-Key with ftch_ API keys from refetch.cloud console
Recommended MCP Servers
1stDibsMCP server for browsing and searching items on 1stDibs marketplace.
2Captcha MCPFull 2Captcha API (43 tools) for Claude Code: 31 captcha solvers + management + webhook.14fetchFetch a URL and get clean Markdown with metadata.
AcrawlAutonomous web crawler.11
AgentfetchToken-budgeted web fetch for AI agents — auto-routes Jina, FireCrawl, Trafilatura, PDF.2AgenticTotem Web ExtractorAI web extraction: send URLs + a JSON Schema, get clean structured data.How it compares
Hosted web fetch and search MCP API, not a local browser-automation skill or self-hosted crawl framework.
FAQ
Who is io.github.lowenbjer/refetch for?
It is for AI agent users who need markdown page fetches, web and news search, and structured extraction without running their own scraper infrastructure.
When should I use io.github.lowenbjer/refetch?
Use it during research and validation when answers must reflect live web content, recent news, or extracted fields from specific URLs.
How do I add io.github.lowenbjer/refetch to my agent?
Create a free API key at https://www.refetch.cloud/console, add the remote server https://refetch.cloud/mcp with streamable-http transport, set the X-API-Key header to your ftch_ key, and reload your MCP client.