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/48

Verdicts · 1,620 servers

✓SAFE913!CAUTION261✕BLOCK446

Transports

A server can list more than one transport.

1,620 servers

1,129–1,152 of 1,620 · ranked by trust score, then stars

kilotest

jrpool/kilotest

!CAUTION C

An ensemble testing service with a focus on accessibility

★ 412 readstreamable-http

uspto_fpd_mcp

Tam1379/uspto_fpd_mcp

!CAUTION C

🚀 Improve patent analysis with a high-performance MCP server for USPTO Final Petition Decisions, featuring context reduction and customizable fields.

★ 41 readstdio

mcp-brasil

ungarbed-triggerfish318/mcp-brasil

!CAUTION C

Conecte agentes de IA a 41 APIs públicas brasileiras para acessar dados do governo, com 363 tools, 87 resources e 62 prompts

★ 29 readstreamable-http

Spotify Scraper

aliakhtari78/spotifyscraper

!CAUTION C

Extract public Spotify data — tracks, albums, artists, playlists, podcasts & lyrics — without the official API. Sync + async, typed models, one dependency.

★ 31628 readstdio

Project OS for Codex

herry2059/project-os-for-codex

!CAUTION C

Open-source control plane for Codex projects: Git-backed context, visible agent progress, scoped MCP access, resumable work, and safe handoffs.

★ 1022 readstdio

Gemini

bsmi021/mcp-gemini-server

!CAUTION C

This project provides a dedicated MCP (Model Context Protocol) server that wraps the @google/genai SDK. It exposes Google's Gemini model capabilities as standard MCP tools, allowing other LLMs (like Cline) or MCP-compatible systems to leverage Gemini's features as a backend workhorse.

★ 3612 readsse

LegalContext

protomated/legal-context-ce

!CAUTION C

LegalContext is an open-source Model Context Protocol (MCP) server that creates a secure, standardized bridge between law firms' document management systems (specifically Clio) and AI assistants (starting with Claude Desktop)

★ 2911 readstdio

MiMo Free API

fu-jie/mimo-free-api-mcp

!CAUTION C

MiMo Free API MCP: 基于官网逆向的高性能 OpenAI / HTTP MCP 代理网关。支持搜索、识图、长文本分段及 SQLite 持久化。

★ 292 readstreamable-http

OpenAgentPack

modelstudioai/OpenAgentPack

!CAUTION C

Manage, review, and migrate cloud AI agents with Git and YAML. The open-source IaC control plane for managed AI agents.

★ 247 readstreamable-http

OpenOSINT

openosint/openosint

!CAUTION C

AI-powered OSINT agent with interactive REPL, MCP server, and CLI. 20 tools. Works with Claude, GPT-4, or local models. For authorized security research only.

★ 1,63110 readstdio

OpenZIM

cameronrye/openzim-mcp

!CAUTION C

OpenZIM MCP is a modern, secure, and high-performance MCP (Model Context Protocol) server that enables AI models to access and search ZIM format knowledge bases offline.

★ 1444 readsse

Code Indexer

zxfgds/mcp-code-indexer

!CAUTION C
★ 348 readstdio

ADK Redis

redis-developer/adk-redis

!CAUTION C

Redis integration for Google Agent Development Kit (ADK) - Memory, Sessions, Search Tools, MCP

★ 302 readstreamable-http

ymm4MCP

SCPgamerscp/ymm4MCP

!CAUTION C

ゆっくりムービーメーカー4のMCPです。将来的にはゆっくり動画を全自動で作れるようにしたいです。

★ 261 readstdio

seo

110kc3/seo

!CAUTION C

Open registry and discovery API for AI products, agent tools, and MCP services.

★ 15 readstreamable-http

figloo

g761007/figloo

!CAUTION C

Chrome extension + local MCP server that lets coding agents like Claude Code and Codex read Figma designs through the web UI with view access: walk layers, inspect properties, take screenshots, export assets. No Figma API; never edits the design.

★ 015 readstdio

Brasil

jxnxts/mcp-brasil

!CAUTION C

MCP Server para 70 APIs públicas brasileiras

★ 1,78610 readstreamable-http

Brasil

mcp-brasil/mcp-brasil

!CAUTION C

MCP Server para 70 APIs públicas brasileiras

★ 1,78310 readstreamable-http

Blockrun Mcp

BlockRunAI/blockrun-mcp

!CAUTION C

Live data for AI agents — search, research, markets, crypto, X/Twitter. Pay-per-call via x402 micropayments.

★ 39119 readstdio

Perplexity

automations-project/vscode-perplexity-mcp

!CAUTION C

Perplexity AI search, reasoning, research, and compute - MCP server, dashboard, and multi-IDE auto-config for VS Code.

★ 7717 readstdio

Nexus

adawalli/nexus

!CAUTION C

MCP Server to make searching openrouter easy

★ 233 readstdio

shyn

shyn-labs/shyn

!CAUTION C

shyn — a register of your days. Local-first memory for your Mac: pages, meetings, notes — encrypted on-device, recalled by your AI over MCP.

★ 16 readstdio

BlockRun

blockrunai/blockrun-mcp

!CAUTION C

Live data for AI agents — search, research, markets, crypto, X/Twitter. Pay-per-call via x402 micropayments.

★ 39519 readstdio

Figma Context

1yhy/figma-context-mcp

!CAUTION C
★ 312 readsse

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.