Everything
Reference / test server with prompts, resources, and tools.
npx -y @modelcontextprotocol/server-everythingSearchable directory of Model Context Protocol servers, filterable by language, transport, and category.
Browse Model Context Protocol servers, filterable by category, language, and transport.
modelcontextprotocol/servers and punkpeye/awesome-mcp-servers. Last update Oct 26, 2026. Reference / test server with prompts, resources, and tools.
npx -y @modelcontextprotocol/server-everythingWeb content fetching and conversion for efficient LLM usage.
npx -y @modelcontextprotocol/server-fetchSecure file operations with configurable access controls.
npx -y @modelcontextprotocol/server-filesystemTools to read, search, and manipulate Git repositories.
npx -y @modelcontextprotocol/server-gitKnowledge graph-based persistent memory system.
npx -y @modelcontextprotocol/server-memoryDynamic and reflective problem-solving through thought sequences.
npx -y @modelcontextprotocol/server-sequential-thinkingTime and timezone conversion capabilities.
npx -y @modelcontextprotocol/server-timeTransparent proxy that sits in front of any other MCP server and shrinks its tool output before it reaches the model. Lossless by default — a denser encoding of the same bytes, with a round-trip gate asserting an exact inverse, so nothing is dropped or summarised. Repeated calls to the same tool emit a delta against the prior result instead of the whole payload. Per-tool opt-in, no decode step, no model in the loop. `uv tool install terse-mcp` then `terse proxy -- <server>`
Checks and validates live x402 payment routes across Base, Solana and Algorand.
Bridge to Moltline Studio's hosted MCP servers: code review, agent governance, business math, regulatory deadlines, crypto tax lots, routing and optimization.
Marketplace of vetted agents with automated security scans plus human review, credential-to-host binding and sandboxed hosted code with default-deny egress.
BotHire agent-to-agent hiring marketplace: search agents and skills by capability and trust score, read live demand, and settle in USDT or USDC across chains.
Edge firewall for exposing internal PostgreSQL databases and APIs to AI agents with strict validation.
Pay-per-call x402 tools (utilities, DNS/WHOIS, blockchain RPC, market and DEX data, prediction markets, security audits) plus an OpenAI-compatible LLM gateway.
Paid crypto intelligence via x402 on Base: preflight checks, trade decisions audited against real prices, risk state, signal history and automation search.
Real-time anomaly detection with NASA-derived sequence mining across blockchain, mempool, stablecoin depeg, aviation and GitHub signals, paid via x402 on Base.
Aggregates multiple MCP servers into one unified server.
Pay-per-call x402 API: geocoding, weather, earthquakes, patents, US case law, Federal Register, scientific papers, DNS/WHOIS, OFAC screening and Census data.
Intelligence endpoints for security, company, threat, compliance, sales, sports, property and health data, paid via x402 micropayments on Base.
Search and call utility APIs for OCR, image, audio, web, math, conversion and geodata, paid per call with x402 USDC on Base.
Search and compare thousands of AI tools, generate step-by-step AI workflows and AI tool stacks by profession from GateOnAI's curated catalog. Free, no API key.
Pay-per-call x402 intelligence APIs on Base and Solana: finance, crypto, legal, immigration, healthcare costs, real estate, tax, climate, sports and science.
Graded MCP registry: live-probes remote servers from the MCP Registry (initialization, tool listing, schema quality, latency, provenance) and publishes grades.
Pay-per-call x402 gateway for agent tools: crypto and DeFi data, web scraping, AI inference, CAPTCHA solving, token security, DNS/WHOIS and gas monitoring.
Daily-verified dataset of LLM API prices for Chinese and global models, with live price queries and token cost estimation.
Search a directory of MCP servers, AI agents, CLI tools and agent skills.
Bridge MCP servers to the agent:// network so other agents can discover and call their tools, with DNS-like identity, mTLS, trust scores and capability search.
Live catalog of x402-payable HTTP tools from AgentBodega: search, public data, social media, status checks, media conversion and agent-readiness checks.
AI agent discovery and marketplace on Nostr with Solana payments (SOL, USDC): NIP-89 discovery, NIP-90 jobs, NIP-44 encryption and on-chain micropayments.
Agent-to-agent (A2A) commerce framework using Nostr NIP-90 and Cashu eCash micro-settlements.
Job aggregator for AI agents across platforms such as x402, RentAHuman and Virtuals.
Job search over employer ATS APIs (Greenhouse, Lever, Ashby, Beisen) with freshness verification, ghost-job scoring and apply links; matching runs client-side.
Self-hostable API gateway and marketplace for AI agents with one API key, REST and streaming access, and health monitoring.
Search, integrate and monetize MCP connectors on the AgentHotspot marketplace.
Search the Remote OpenClaw directory of MCP servers, agent skills and plugins for names, links and install commands; also available as a CLI.
Skill marketplace where AI agents discover, purchase and integrate skills.
AnyAPI: scraping and data APIs for social media, search results, Google Maps, e-commerce and web data, with normalized JSON and automatic provider failover.
CLIRank API intelligence for coding agents: discover, compare and get recommendations for APIs scored on agent-friendliness.
Turn any REST API into a hosted MCP server, with ready-made public servers for GitHub, Stripe, Slack, OpenAI, Notion and more.
AIMarket Hub plugin that packages AIMarket capabilities as self-hosted MCP servers with a Docker image, manifest and client config.
Local MCP proxy that merges downstream servers behind a few meta-tools with progressive discovery and compresses large tool outputs, with full output on demand.
Search, inspect and run data APIs for SEO and AI visibility, finance, social, web search, sales and agent mail through a few meta-tools, with spending caps.
Self-hosted MCP gateway fronting upstream servers with per-tool access control, guardrails, rate limits and usage logs; also proxies LLM and A2A agent traffic.
Live web data as clean JSON: search engines, social media, e-commerce (Amazon, eBay, Etsy), real estate, jobs, travel, news, finance and company intel.
MeshMarket agent-to-agent capability exchange: agents browse, self-onboard, rent memory, reasoning and safety per call, and publish their own tools.
Agent-to-agent marketplace where AI agents discover, invoke and pay for other agents' services using USDC on Base.
DeepSeek AI chat and reasoning with multi-turn sessions, function calling, thinking mode and cost tracking.
Build, validate and deploy multi-agent AI solutions on the ADAS platform: design skills with tools and manage the solution lifecycle.
Meta-MCP server that unifies AI coding assistants (Codex, Claude Code, Cursor, Gemini) behind one interface with automatic discovery and zero configuration.
Search the awesome-mcp.tools catalog of MCP servers from an MCP client or the terminal.
Multi-model AI media generation (image, video, audio, face swap, talking avatars) plus chat with models such as Claude, GPT, Gemini and DeepSeek.
Self-hosted MCP for teams: deploy your own Python tools from git to Kubernetes (AKS/GKE/EKS/local) with canary rollouts, OAuth sign-in, autoscaling, Nvidia NeMo guardrails, roles per group, secrets and an audit log of every call.
Access AI models such as GPT-5, Claude, Gemini, Grok and DeepSeek, paid per use via x402 micropayments in USDC on Base.
Pay-per-request AI chat completions across supported models via x402, with automatic Solana USDC payments and fail-closed spending caps.
Provenance gateway for stdio MCP servers that blocks tool calls whose destination or content came from an untrusted tool result, with schema-derived policies.
x402 discovery hub: search engine for agent services, paid with USDC on Base via x402.
Browse LLMs from OpenAI, Anthropic, Google, Qwen, DeepSeek and others with live prices, and route chat completions through the OrcaRouter gateway with fallback.
Unified MCP platform that connects many apps, tools and services behind one interface.
Dependency graphs for TypeScript, JavaScript, Python, Go, Rust and C, with arc diagram visualization, health scoring, dead code detection and a temporal graph.
Self-hosted MCP server registry for enterprise AI agents.
The Model Context Protocol (MCP) is an open standard that lets AI assistants like Claude connect to external tools, data sources, and services through a common interface. An MCP server is a small program that exposes capabilities (file access, database queries, API calls, web search) to an AI client. Instead of writing a custom integration for every tool, MCP gives any AI agent a uniform way to discover and use them.
This MCP Server Directory is a searchable, filterable browser for over 4,000 MCP servers, combining the official MCP registry with community servers. It is auto-refreshed every week directly from the upstream repository, so you always see what is current without scrolling through a long markdown README.
A weekly GitHub Action fetches two upstream lists, the official modelcontextprotocol/servers README and the community punkpeye/awesome-mcp-servers list, parses the server entries and commits any changes directly into this site's data file. Newly added servers appear here within 7 days without manual maintenance. Note that neither list is the official MCP Registry, so a server published only there will not show up yet. Spot something wrong or missing? The Suggest update by email button at the bottom opens a pre-filled message.
The official list is a single long markdown file. This directory adds the layer that is missing: full-text search, faceted filters, copyable install commands, archived-server detection, and a clear official vs community badge. All of it runs entirely in your browser with no tracking or sign-up.
What is an MCP server?
An MCP server is a small program that exposes tools, data sources, or services (file access, database queries, web search, API calls) to AI assistants like Claude through the Model Context Protocol, an open standard. Instead of a custom integration per tool, any MCP-compatible client can discover and use them through one interface.
How do I install an MCP server?
Most MCP servers install with a single command, typically npx for TypeScript servers or uvx/pip for Python servers. Each entry in this directory shows its install command with a one-click copy button, plus a link to the source repository with full configuration instructions.
How many MCP servers exist?
This directory currently tracks over 4,000 MCP servers, combining the official modelcontextprotocol registry with community servers. The list is re-synced automatically every week, so newly published servers appear within 7 days.
What is the difference between official and community MCP servers?
Official servers are maintained by the Model Context Protocol team or by the company behind the integrated service, and are generally more stable. Community servers are third-party contributions with varying maintenance levels. This directory labels each server and lets you filter to official servers only.
Free forever, no ads, no tracking. Support the project