Evaluation record · mcp-server-fetch

MCP Fetch Server

v2026.6.4

Anthropic

MCPhttpapimcpmodel-context-protocol

Strong

About This MCP

Official MCP reference server for fetching web content and converting HTML to markdown. Enables AI models to retrieve and process web pages, documentation, and online resources. One of the seven reference servers that remain actively maintained after the 2025-05-29 archival of non-core servers; MCP governance moved to the Agentic AI Foundation under the Linux Foundation on 2025-12-09 (latest stable spec 2025-11-25; 2026-07-28 spec revision at release-candidate stage as of 2026-07).

Last Evaluated: July 9, 2026

Official Website

Trust Vector Analysis

Dimension Breakdown

🚀Performance & Reliability

fetch success rate

Fetch operation success testing

Evidence

MCP Fetch Server — Reliable HTTP fetching with standard Node.js/Python libraries

highVerified: 2026-07-09

html to markdown accuracy

Conversion quality assessment

Evidence

Turndown Library — Uses Turndown for HTML-to-Markdown conversion with high fidelity

highVerified: 2026-07-09

timeout handling

Timeout behavior testing

Evidence

Implementation Review — Configurable timeouts with graceful failure handling

mediumVerified: 2026-07-09

rate limit handling

Rate limiting behavior analysis

Evidence

HTTP Client Implementation — Respects robots.txt and implements basic rate limiting

mediumVerified: 2026-07-09

error recovery

Error handling testing

Evidence

MCP Implementation — Handles HTTP errors, redirects, and network failures gracefully

mediumVerified: 2026-07-09

🛡️Security

ssrf protection

SSRF vulnerability testing

Evidence

Security Analysis — Limited SSRF protections; can fetch internal URLs if allowed

highVerified: 2026-07-09

url validation

Input validation testing

Evidence

Implementation Review — Basic URL validation but no blocklist for internal networks

mediumVerified: 2026-07-09

ssl certificate validation

SSL/TLS security review

Evidence

HTTPS Implementation — Validates SSL certificates by default

highVerified: 2026-07-09

malicious content protection

Content security assessment

Evidence

Security Analysis — No built-in scanning for malicious JavaScript or content

mediumVerified: 2026-07-09

sensitive data exposure

Data flow security analysis

Evidence

MCP Security Model — Fetched content sent directly to LLM provider without filtering

highVerified: 2026-07-09

🔒Privacy & Compliance

web content exposure

Data flow analysis

Evidence

MCP Data Flow — Fetched web content transmitted to LLM provider for processing

highVerified: 2026-07-09

user agent disclosure

Privacy disclosure review

Evidence

HTTP Headers — Sends identifiable User-Agent header to visited sites

mediumVerified: 2026-07-09

cookie handling

Cookie management assessment

Evidence

Implementation Review — Does not persist cookies; stateless fetching only

mediumVerified: 2026-07-09

third party data sharing

Data sharing analysis

Evidence

LLM Provider Policies — All fetched content shared with LLM provider per their privacy policy

highVerified: 2026-07-09

paywalled content respect

Content access policy review

Evidence

Implementation Review — No authentication support; cannot bypass paywalls ethically

mediumVerified: 2026-07-09

👁️Trust & Transparency

documentation quality

Documentation completeness review

Evidence

MCP Fetch Docs — Comprehensive documentation with usage examples and configuration

highVerified: 2026-07-09

operation visibility

Logging and traceability assessment

Evidence

MCP Protocol — All fetch operations logged in MCP transaction logs

highVerified: 2026-07-09

open source transparency

Source code review

Evidence

GitHub Repository — Fully open source implementation with MIT license

highVerified: 2026-07-09

security disclosure

Security documentation review

Evidence

MCP Security Documentation — Clear disclosure of SSRF risks and data sharing implications

mediumVerified: 2026-07-09

⚙️Operational Excellence

ease of setup

Setup complexity assessment

Evidence

MCP Setup Guide — Simple setup with no authentication required for basic usage

highVerified: 2026-07-09

fetch performance

Performance benchmarking

Evidence

Performance Testing — Performance depends on target website; typically 500ms-3s per fetch

mediumVerified: 2026-07-09

reliability

Uptime and stability analysis

Evidence

Implementation Analysis — Stable implementation with minimal dependencies

highVerified: 2026-07-09

conversion quality

Conversion accuracy testing

Evidence

Turndown Library — High-quality HTML to Markdown conversion preserving structure

highVerified: 2026-07-09

community adoption

Community activity analysis

Evidence

GitHub Community — Widely adopted as core MCP server for web content access

Anthropic - Donating MCP to the Agentic AI Foundation — fetch is one of the seven reference servers still actively maintained after the 2025-05-29 archival; MCP governance moved to the Agentic AI Foundation under the Linux Foundation on 2025-12-09

PyPI - mcp-server-fetch — Latest release 2026.6.4 (published 2026-06-04); still listed among the seven maintained reference servers in modelcontextprotocol/servers README (repo ~88,300 stars, not archived)

mediumVerified: 2026-07-09

Strengths

+Simple, zero-configuration setup for basic web content fetching
+High-quality HTML to Markdown conversion using Turndown library
+Open source and actively maintained under the MCP project (Agentic AI Foundation)
+Reliable HTTP fetching with timeout and error handling
+Respects robots.txt and implements ethical web scraping practices
+Excellent documentation and integration with MCP ecosystem

Limitations

!Limited SSRF protection; can potentially access internal networks
!All fetched content exposed to LLM provider without filtering
!No built-in authentication for paywalled or restricted content
!No malicious content scanning or sanitization
!Performance dependent on target website responsiveness
!Basic rate limiting may cause issues with aggressive scraping
!STATUS 2026-07-09: actively maintained reference server (one of 7 retained after the 2025-05-29 archival); latest PyPI release 2026.6.4 (2026-06-04); no CVEs against this reference implementation (note: CVE-2025-65513 SSRF affects the unrelated third-party npm package 'mcp-fetch-server', not this Python mcp-server-fetch); governance under the Agentic AI Foundation (Linux Foundation, 2025-12-09)

Metadata

license: MIT

supported platforms

0: All platforms with Node.js/Python

programming languages

0: TypeScript

1: Python

mcp version: 1.0

github repo: https://github.com/modelcontextprotocol/servers

github stars: 88259

api dependency: HTTP/HTTPS, Turndown

authentication: None required

first release: 2024-11

maintained by: MCP project (Agentic AI Foundation / Linux Foundation since 2025-12-09)

status: Official - Active

transport types

0: stdio

installation methods

0: pip

Use Case Ratings

code generation

Useful for fetching documentation and code examples from the web

customer support

Good for retrieving knowledge base articles and support documentation

content creation

Excellent for research, gathering sources, and content aggregation

data analysis

Useful for fetching data from web sources, but not optimized for structured data

research assistant

Ideal for academic research, gathering sources, and documentation retrieval

legal compliance

Limited applicability; risk of exposing confidential research to LLM providers

healthcare

Low suitability due to privacy concerns with medical content

financial analysis

Moderate risk; requires careful consideration of data sensitivity

education

Excellent for educational research and accessing learning materials

creative writing

Great for research, inspiration, and gathering reference materials