MCP servers · least privilege

Read-only MCP servers

1,620 MCP servers declare tools that only read: none of their tools is classed write or destructive. That makes them the safest place to start giving an agent access. Read-only describes the tools, not the code, so the audit verdict still decides whether to connect one.

Ask an AI about this pageGChatGPTCClaudePPerplexityGeGeminiXGrok
More
Cite

Cite this page

ModelsAtlas. “Read-only MCP servers”. Retrieved 29 Oct 2026. https://modelsatlas.com/mcp/read-only/page/2

Verdicts · 1,620 servers

✓SAFE913!CAUTION261✕BLOCK446

Transports

A server can list more than one transport.

1,620 servers

25–48 of 1,620 · ranked by trust score, then stars

Deepwiki

regenrek/deepwiki-mcp

✓SAFE B

📖 MCP server for fetch deepwiki.com and get latest knowledge in Cursor and other Code Editors

★ 1,3862 readsse

mcp-deepwiki

regenrek/deepwiki-mcp

✓SAFE B

📖 MCP server for fetch deepwiki.com and get latest knowledge in Cursor and other Code Editors

★ 1,3852 readsse

Claude Code

steipete/claude-code-mcp

!CAUTION B

Claude Code as one-shot MCP server to have an agent in your agent.

★ 1,3141 readstdio

Quickstart Resources

modelcontextprotocol/quickstart-resources

✓SAFE B

A repository of servers and clients from the Model Context Protocol tutorials

★ 1,2054 readstdio

Ref Tools

ref-tools/ref-tools-mcp

✓SAFE B

Helping coding agents never make mistakes working with public or private libraries without wasting the context window.

★ 1,1753 readstdio

RedNote

ifuryst/rednote-mcp

✓SAFE B

🚀MCP server for accessing RedNote(XiaoHongShu, xhs).

★ 1,1114 readstdio

Lightswind UI

codewithmuhilan/lightswind-ui-library

✓SAFE B

The AI-native CLI-first React component library. 160+ animated, accessible, production-ready components with MCP Server support for Cursor, Claude & GitHub Copilot. Copy-paste architecture — you own the code.

★ 1,0788 readstdio

Minima

dmayboroda/minima

✓SAFE B

On-premises conversational RAG with configurable containers

★ 1,0491 readstdio

Mcpdoc

langchain-ai/mcpdoc

✓SAFE B

Expose llms-txt to IDEs for development

★ 1,0343 readstdio

Chatsum

chatmcp/mcp-server-chatsum

✓SAFE B

Query and Summarize your chat messages.

★ 1,0271 readstdio

Nuxt

antfu/nuxt-mcp

✓SAFE B

MCP server helping models to understand your Vite/Nuxt app better.

★ 9117 readsse

Nuxt Dev

antfu/nuxt-mcp-dev

✓SAFE B

MCP server helping models to understand your Vite/Nuxt app better.

★ 9117 readsse

Openapi

janwilmake/openapi-mcp-server

✓SAFE B

Allow AI to wade through complex OpenAPIs using Simple Language

★ 9042 read

WebMCP

jasonjmcghee/webmcp

!CAUTION B

Early WebMCP proposal / implementation - since evolved and worked on by much more capable folks that develop the web: https://github.com/webmachinelearning/webmcp

★ 8672 readstdio

Browser Use

co-browser/browser-use-mcp-server

✓SAFE B

Browse the web, directly from Cursor etc.

★ 8442 read

Mcp-Nixos

utensils/mcp-nixos

✓SAFE B

MCP-NixOS - Model Context Protocol Server for NixOS resources

★ 8441 read

Fetch

zcaceres/fetch-mcp

✓SAFE B

A flexible HTTP fetching Model Context Protocol server.

★ 8276 readstdio

mcp-fetch-server

zcaceres/fetch-mcp

✓SAFE B

A flexible HTTP fetching Model Context Protocol server.

★ 8266 readstdio

Idea Reality

mnemox-ai/idea-reality-mcp

✓SAFE B

Pre-build reality check for AI coding agents. Scans GitHub, HN, npm, PyPI, Product Hunt. MCP server. 290+ stars.

★ 8221 readstdio

Acemcp

qy527145/acemcp

!CAUTION B

一个将ACE(Augment Context Engine) 做成MCP的项目

★ 8031 readstdio

OpenBrand

ethanjyx/openbrand

!CAUTION B

Extract brand assets (logos, colors, backdrops) from any website; open source alternative to Brand.dev

★ 7991 readstdio

Claude ChatGPT

syedazharmbnr1/claude-chatgpt-mcp

✓SAFE B

A Claude MCP tool to interact with the ChatGPT desktop app on macOS

★ 7931 readstdio

How To Cook

worryzyy/howtocook-mcp

✓SAFE B

基于Anduin2017 / HowToCook (程序员在家做饭指南)的mcp server

★ 7735 readsse

obsidian-mcp

StevenStavrakis/obsidian-mcp

✓SAFE B

A simple MCP server for Obsidian

★ 7412 readstdio

Other risk hubs

By job

Questions

What does read-only mean here?
A server is listed here when its source declares at least one tool and every tool is classed read: no write tools and no destructive tools.
Is a read-only server always safe?
No. 913 are SAFE, 261 CAUTION and 446 BLOCK on this list. A tool list says what the tools do, not whether the surrounding code is clean, so a read-only server can still be BLOCK.
Why are some servers on no risk list?
615 servers declare no tools we can read, because they build the list at runtime. They cannot be called read-only, so they appear on none of the risk hubs.

Keep going in your AI

Opens with this page as the source. We add nothing else to the prompt.