Fetch
Fetches web pages and converts them to markdown for efficient LLM use.
How it runs
This server runs on your own machine. Your MCP client starts it as a local process (usually through npx, uvx or Docker) and talks to it over standard input/output. It runs with your user's permissions, so scope paths and credentials tightly.
Who maintains it
This is a reference server from the Model Context Protocol project. Reference servers demonstrate MCP features and are a good way to learn the protocol; check whether they suit production use.
Safety checklist
- Review which tools it exposes, and prefer read-only modes where available.
- Use scoped, least-privilege credentials or a test account first.
- Keep human approval on for actions with side effects (writes, payments, deletions).
- Treat content returned from the web or user data as untrusted, because it can contain prompt-injection attempts.
Similar MCP servers
Filesystem
Secure file operations with configurable access controls.
Git
Read, search and manipulate local Git repositories.
Memory
Knowledge-graph-based persistent memory for agents.
Sequential Thinking
Structured, reflective step-by-step problem solving.
Time
Current time and time-zone conversions.
Everything
Test server exercising every MCP feature: prompts, resources and tools.