MCP servers · connect in Claude Desktop

Claude Desktop MCP servers

3,432 MCP servers have a connect command for Claude Desktop. 1,660 are SAFE, 638 CAUTION and 1,134 BLOCK. Every one was read in full before it was listed.

Ask an AI about this pageGChatGPTCClaudePPerplexityGeGeminiXGrok
More
Cite

Cite this page

ModelsAtlas. “Claude Desktop MCP servers”. Retrieved 29 Oct 2026. https://modelsatlas.com/mcp/for/claude-desktop/page/143

Verdicts · 3,432 servers

✓SAFE1,660!CAUTION638✕BLOCK1,134

Transports

A server can list more than one transport.

Tool risk

1,009

servers expose at least one destructive tool: delete, overwrite or run. The median server with a readable tool list exposes 9 tools.

Servers with no destructive tools →

3,432 servers for Claude Desktop

3,409–3,432 of 3,432 · ranked by trust score, then stars

agent-nuvira

imdheerajKube/agent-nuvira

✕BLOCK F
★ 09 destructivestdio

knowledge-graph-system

aaronsb/knowledge-graph-system

✕BLOCK F

Kappa Graph — κ(G). A semantic knowledge graph where knowledge has weight. Extracts concepts, measures grounding strength, preserves disagreement, traces everything to source.

★ 1282 destructivestdio

VibeShell

veithly/vibeshell

✕BLOCK F

Local-first SSH/SFTP workspace for humans and coding agents: visible operations, shared sessions, integrated files and discoverable plugins.

★ 8420 read

hackmyagent

opena2a-org/hackmyagent

✕BLOCK F

Metasploit for AI agents: scan, attack, and fix AI agents and MCP servers. Open source security toolkit.

★ 831 writestdio

cybara

metaspartan/cybara

✕BLOCK F

Cybara - Your agents. Your tools. Your runtime. Harness the power of AI

★ 314 destructivestreamable-http

GraphFlow

Roarpeng/GraphFlow

✕BLOCK F

Local-first code knowledge graph and context harness for coding agents. MCP + DeepSeek Harness (dsh) plugin.

★ 194 writestdio

Apx

agentprojectcontext/apx

✕BLOCK F

Meet your crew. APX is an open-source Agent OS for running AI agents on your own machine — memory in plain files, agents that work together, Claude Code, Codex, Telegram, WhatsApp and MCP.

★ 164 destructivestdio

AgenticOS-Workbench

zzoretich/AgenticOS-Workbench

✕BLOCK F

UniDeX Agent Harness: a second brain and supervised Chief of Staff for Claude Code and Codex CLI, with persistent memory, recall and session capture, plus proposals you approve before anything changes. Lives in a plain-Markdown vault, with the UniDeX macOS app as its live dashboard.

★ 32 writestdio

cli

getmcpm/cli

✕BLOCK F

MCP security guard + package manager. Block prompt injection, tool poisoning and rug-pulls in Model Context Protocol servers at runtime; trust-scored installs with Sigstore provenance for Claude Desktop, Claude Code, Cursor, VS Code, Windsurf and Gemini CLI. Local, deterministic, no LLM in the loop.

★ 31 destructivestdio

exceptd Security

blamejs/exceptd-skills

✕BLOCK F

AI security skills grounded in mid-2026 threat reality, not stale framework documentation

★ 011 read

great-cto

avelikiy/great_cto

✕BLOCK F

You already have the agent. This is everything around it. great_cto runs Claude Code as a pipeline of 70 specialist agents — an independent model checks each stage before the next builds on it, spending caps refuse rather than warn, and three decisions stay yours: what gets built, how, and whether i

★ 1039 read

bv-mcp

MadaBurns/bv-mcp

✕BLOCK F

Open-source DNS & email security scanner. One MCP endpoint, 57 checks, zero install. Cloudflare Workers.

★ 99 readstreamable-http

tribunal-kit

Harmitx7/tribunal-kit

✕BLOCK F

A universal, native-Rust governance layer for AI coding agents. Enforces Subagent-Driven Development (SDD) to stop hallucinations before they hit disk.

★ 51 destructive

BreachPilot

braydos-h/BreachPilot

✕BLOCK F

Autonomous Offensive Security Agent for Red Team Operations

★ 48 writestdio

sequant

sequant-io/sequant

✕BLOCK F

AI coding agent orchestrator — resolve GitHub issues end-to-end with isolated git worktrees, quality gates, and an MCP server. Works with Claude Code or Aider.

★ 11 writesse

xavier

iberi22/xavier

✕BLOCK F

Xavier — Rust vector memory & knowledge graph for AI agents (HTTP/CLI/MCP). Powers SWAL communal context

★ 11 destructive

Jarela

CircuitWall/jarela

✕BLOCK F

Jarela is a lightweight, full‐featured multi‐agent application that runs locally on your machine.

★ 029 destructivestreamable-http

Pollinations

pollinations/pollinations

✕BLOCK F

Your Friendly Open-Source Gen-AI Platform

★ 5,1011 writestreamable-http

🐋 DeepSeek CLI

sluisr/deepseek-cli

✕BLOCK F

DeepSeek CLI — unofficial adaptation of Gemini CLI by sluisr. Full agentic AI in your terminal using DeepSeek API.

★ 52 destructivestdio

bce-engine

blueprint-conformance/bce

✕BLOCK F

Blueprints with Teeth — the blueprint conformance engine: authored architectural intent, enforced in CI.

★ 13 write

Seerai

dralkh/seerai

✕BLOCK F

Zotero AI plugin Research assistant for Zotero 10. Chat with your library, run federated scholarly search, RAG, OCR, systematic reviews, and manage cloud storage. Includes standalone MCP, Agentic capabilities, and skills library.

★ 882 destructivestdio

llm-trust-guard

nkratk/llm-trust-guard

✕BLOCK F
★ 214 writestreamable-http

Gateway Registry

agentic-community/mcp-gateway-registry

✕BLOCK F

Enterprise-ready MCP Gateway & Registry that centralizes AI development tools with secure OAuth authentication, dynamic tool discovery, and unified access for both autonomous AI agents and AI coding assistants. Transform scattered MCP server chaos into governed, auditable tool access with Keycloak/E

★ 9442 writestdio

✦ Mipham Code

One-Mipham/mipham-code

✕BLOCK F
★ 02 destructive

More about Claude Desktop

Other client hubs

By risk

Questions

How do I add an MCP server to Claude Desktop?
Paste the JSON block shown on the server's page into the mcpServers section of Claude Desktop's configuration file.
Which Claude Desktop MCP servers are safe?
Of the 3,432 servers with a Claude Desktop connect command, 1,660 are SAFE, 638 CAUTION and 1,134 BLOCK. Cards are ranked by trust score, so the safest are first; open one to read its findings before you connect it.
Are any of these servers remote?
1,261 of the 3,432 list a remote transport (streamable HTTP or SSE). The rest run as a local process your client starts.

Keep going in your AI

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