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/35

Verdicts · 1,620 servers

✓SAFE913!CAUTION261✕BLOCK446

Transports

A server can list more than one transport.

1,620 servers

817–840 of 1,620 · ranked by trust score, then stars

danielortiz0220-nyc-restrooms

danielortiz0220/nyc-restroom-mcp

✓SAFE B

Find NYC public restrooms and their latest Parks inspection status, from NYC Open Data

★ 12 readstdio

faceit-mcp

saivi3171/faceit-mcp

✓SAFE B

Connect Claude to live CS2 FACEIT data to retrieve player stats, match history, ELO trends, and head-to-head comparisons via the Model Context Protocol.

★ 18 read

gfc

jjjorfi/gfc

✓SAFE B

git-for-context — LLM context dedup: unchanged file re-reads collapse to hash-verified stubs, changed files return diffs. OpenCode plugin + zero-dep MCP server. Measured −34% per-turn context.

★ 11 read

goaimoat-ai-visibility-mcp

jayniebingyu-cyber/goaimoat-ai-visibility-mcp

!CAUTION B

GoAI Moat - AI Visibility Audit MCP server. Diagnose a brand visibility in AI answers (ChatGPT, Perplexity, AI Overviews). https://mcp.goaimoat.com/mcp

★ 13 readstreamable-http

graphmaxer-gw1

Graphmaxer/gw1-mcp

✓SAFE B

GW1 build compiler MCP server: skill data, template code encode/decode, validation, hero roster. Read-only.

★ 12 readstdio

internet-radio-mcp

AlonDrilich/internet-radio-mcp

!CAUTION B

MCP server for internet radio: search 50,000+ stations, get stream URLs, top/trending stations, countries and genres (Radio Browser public-domain data). By 72FM.

★ 15 readstdio

kazkozdev-abscissa

KazKozDev/abscissa

✓SAFE B

A safety-aware MCP server that lets AI agents manage Linear issues, projects, cycles, and dependencies.

★ 117 readstdio

local-terminal-mcp

luckysolanki902/local-terminal-mcp

✓SAFE B

Run local commands from ChatGPT or Claude via a safe, allowlisted MCP server. Work on your codebase and generate images/sprites/assets locally from ChatGPT without spending Codex/API tokens. stdio for Claude Desktop/Code; capability-URL or bearer auth over HTTP for ChatGPT via ngrok or cloudflared.

★ 13 readstdio

machinegrade-validate

machinegrade/validate

✓SAFE B

Deterministic validation for AI-generated artifacts: JSON Schema, OpenAPI response, SQL. Typed verdicts with fix hints. Metered API + MCP.

★ 12 readstdio

MCP-Ambari-API

call518/MCP-Ambari-API

✓SAFE B

🚀 Natural-language Hadoop cluster management via MCP. Connect LLM agents (Claude, GPT, etc.) directly to Apache Ambari REST API — control services, inspect configs, query metrics, and manage alerts with AI. Built with FastMCP & aiohttp.

★ 11 readstdio

mcp-atlas

urbankitstudio/mcp-atlas

✓SAFE B

MCP server: find a US county's official parcel ArcGIS endpoint and search it by owner name or APN. No API key. npx -y @urbankitstudio/mcp-atlas

★ 14 readstdio

mcp-marmiton

smeet666/mcp-marmiton

✓SAFE B

MCP server for Marmiton: search French recipes and scale ingredient quantities to any number of servings. No API key.

★ 13 readstdio

mcp-server

TranscriptFetch/mcp-server

✓SAFE B

MCP server for TranscriptFetch: transcripts from YouTube, TikTok, Instagram and media URLs, with AI transcription when captions are missing, plus YouTube search, channel and playlist listing, from any MCP client.

★ 14 readstdio

mcp-server-axiom

ThetaBird/mcp-server-axiom-js

✓SAFE B
★ 13 readstdio

MemAI

Filipe-Soares-de-Almeida/MemAI

!CAUTION B

Long-term memory MCP server for AI agents — one ACID SQLite file, FTS5 BM25 retrieval, nested domains, diagrams and a local curation dashboard

★ 12 read

mohsinmshabbir-datasaudi

mohsinmshabbir/datasaudi-mcp

✓SAFE B

Saudi economic data for Claude, Cursor & any MCP client — English or Arabic, keyless. Refuses to silently mislead.

★ 13 readstdio

moodle-mcp-server

csmediapro/moodle-mcp-server

✓SAFE B

Open-source MCP server that connects AI agents to Moodle LMS data through Moodle Web Services.

★ 18 readstdio

orion-celestia-intel

orioninfra/orion-agent-kit

!CAUTION B

Celestia network intelligence for AI agents: MCP server over Orion's daily analysis. Read-only.

★ 111 readstdio

PaperCortex

chetanmorey1/PaperCortex

✓SAFE B

Organize and search Paperless-ngx documents with local AI, auto-classification, receipt extraction, and accounting export.

★ 15 readsse

propline-mcp

proplineapi/propline-mcp

✓SAFE B

Live odds, cross-book +EV and graded player-prop results across 37 books. Hosted endpoint included.

★ 130 readstdio

scopewalker-mcp

timohaa/scopewalker-mcp

✓SAFE B

Codebase analysis MCP server for AI agents: complexity metrics, size thresholds, prop drilling, and code smell detection built on tree-sitter and tokei.

★ 19 readstdio

songcheck

afghanfansmedia-ai/songcheck-mcp

✓SAFE B

Is this song AI or human? Detect AI-generated music (Suno, Udio, and more) from any AI agent. MCP server by Khaled Media.

★ 14 read

sovantica-engrava

sovantica/engrava-mcp

✓SAFE B

Model Context Protocol server for Engrava — expose an AI agent memory database to any MCP client (Claude Desktop, Cursor, VS Code...) over stdio.

★ 11 readstdio

substack-mcp-cli

thenavidm/substack-mcp-cli

✓SAFE B

Substack MCP server and CLI for Claude Code, Codex and AI agents. 65 tools for drafts, posts, Notes, subscribers, analytics, tags, comments, and researching other writers.

★ 14 read

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.