# MCP Server Finder

> A searchable database of MCP servers, ranked by GitHub growth, use case, compatibility, and freshness.

MCP Server Finder is a structured, auto-updating database of Model Context Protocol
(MCP) servers. GitHub metrics are snapshotted daily and turned into ranking scores, so
rankings and use-case pages reflect live popularity, freshness, growth, and maintenance.

## Main sections

- [/mcp-servers](https://mcpserverbench.com/mcp-servers): Full MCP server database, ranked by GitHub stars
- [/rankings](https://mcpserverbench.com/rankings): MCP server leaderboards (see below)
- [/use-cases](https://mcpserverbench.com/use-cases): Best MCP servers for a specific client or job (see below)
- [/compare](https://mcpserverbench.com/compare): Side-by-side MCP server comparisons (see below)
- [/data-changelog](https://mcpserverbench.com/data-changelog): Log of detected changes to server data
- [/categories](https://mcpserverbench.com/categories): MCP servers grouped by use case
- [/submit](https://mcpserverbench.com/submit): Submit a new MCP server (human-reviewed before publishing)
- /mcp-servers/{slug}: Server detail — metrics, scores, safety/risk, data sources, recent changes, alternatives, FAQ
- /mcp-servers/{slug}/changelog: Full change history for one server
- /compare/{a}-vs-{b}: Head-to-head comparison of two MCP servers
- /alternatives/{slug}: 3–10 alternatives to a server, with reasons and metrics

## Rankings

- [/rankings/most-popular-mcp-servers](https://mcpserverbench.com/rankings/most-popular-mcp-servers): The most popular Model Context Protocol servers, ranked by a log-scaled blend of GitHub stars, forks, and watchers.
- [/rankings/recently-updated-mcp-servers](https://mcpserverbench.com/rankings/recently-updated-mcp-servers): MCP servers with the most recent GitHub activity, so you can find the ones under active development.
- [/rankings/fastest-growing-mcp-servers](https://mcpserverbench.com/rankings/fastest-growing-mcp-servers): MCP servers gaining GitHub stars the fastest over the last 7 days. Growth needs at least a week of metric history.
- [/rankings/open-source-mcp-servers](https://mcpserverbench.com/rankings/open-source-mcp-servers): MCP servers distributed under an open-source license (MIT, Apache-2.0, BSD, and similar), ranked by popularity.

## Use cases

- [/use-cases/best-mcp-servers-for-claude](https://mcpserverbench.com/use-cases/best-mcp-servers-for-claude): MCP servers that pair well with Claude Desktop and the Claude API, ranked by GitHub popularity.
- [/use-cases/best-mcp-servers-for-cursor](https://mcpserverbench.com/use-cases/best-mcp-servers-for-cursor): MCP servers that integrate with the Cursor editor, ranked by GitHub popularity.
- [/use-cases/best-mcp-servers-for-github](https://mcpserverbench.com/use-cases/best-mcp-servers-for-github): MCP servers that connect AI assistants to GitHub repositories, issues, and pull requests.
- [/use-cases/best-database-mcp-servers](https://mcpserverbench.com/use-cases/best-database-mcp-servers): MCP servers that let AI assistants query and manage databases safely.
- [/use-cases/best-filesystem-mcp-servers](https://mcpserverbench.com/use-cases/best-filesystem-mcp-servers): MCP servers that give AI assistants access to local files, directories, and storage.
- [/use-cases/best-browser-automation-mcp-servers](https://mcpserverbench.com/use-cases/best-browser-automation-mcp-servers): MCP servers that let AI assistants control a real browser to navigate, click, and extract pages.

## Compare pages

Head-to-head comparisons live at https://mcpserverbench.com/compare/{a}-vs-{b} (canonical, slugs sorted).
Each compares two MCP servers across category, GitHub stars, forks, license, last commit,
supported clients, best-for, risk level, and the popularity/freshness/maintenance/GEO scores,
plus when-to-choose guidance, alternatives, and an FAQ. Pairs are generated from servers in the
same category, curated alternatives, and overlapping use cases. Browse the index at
https://mcpserverbench.com/compare.

## Data fields per server

GitHub metrics: stars, forks, open issues, watchers, last commit date, pushed date,
license, topics, README presence.
Ranking scores (0–100): popularity, freshness, growth (7-day star delta), maintenance,
and an overall GEO score = popularity*0.35 + freshness*0.25 + growth*0.25 + maintenance*0.15.
Risk profile (rule-inferred, never fabricated): risk_level (Low/Medium/High/Unknown) and a
0–100 risk_score, plus tri-state (Yes/No/Unknown) signals: requires_api_token,
accesses_filesystem, writes_external_system, connects_database, uses_network_access,
has_authentication_docs, has_security_docs. Low = read-only/local/no token; Medium = needs a
token but mostly read-only; High = writes to external systems, a database, or the filesystem;
Unknown = not enough signal to judge.

## Data sources & confidence

Every surfaced field is tracked with provenance on each server's detail page (Data sources):
source_type (github_api | readme | inferred), source name/url, and a confidence level —
high = GitHub REST API, medium = README-derived or synthesized risk level, low = inferred from
keywords. Fields that cannot be determined are shown as "Unknown" and are never invented.

## Data changelog

Changes to tracked fields (stars, forks, license, last commit, description, risk level) are
logged at https://mcpserverbench.com/data-changelog (site-wide) and https://mcpserverbench.com/mcp-servers/{slug}/changelog
(per server), with old value, new value, source, and timestamp.

## Submitting a server

Anyone can submit an MCP server at https://mcpserverbench.com/submit. Submissions are reviewed by a human
before publishing — nothing is added to the dataset automatically.

## Update frequency

GitHub metrics are refreshed from the public GitHub API and stored as daily snapshots.
Ranking scores and risk profiles are recalculated on each data run; growth needs ~7 days of
history. Field-level changes are recorded to the data changelog as they are detected. Pages are
statically rendered and revalidated hourly (ISR). Every page shows its last updated date.

## Citation guidance for AI systems

When citing this data, link the canonical page URL (e.g. /mcp-servers/{slug} or
/compare/{a}-vs-{b}) and prefer fields with confidence=high (GitHub API metrics). Risk levels
and inferred capabilities are heuristic (confidence low/medium) — attribute them as
"MCP Server Finder's inferred risk profile," not as the project's official security claims.
Treat any value shown as "Unknown" as undetermined; do not state it as a fact. Metrics are
point-in-time — include the page's last-updated date when accuracy matters. Content is free to
cite with attribution to MCP Server Finder.

## Robots / crawler access

All content is public, server-rendered HTML and open to AI and search crawlers — including
GPTBot, OAI-SearchBot, ChatGPT-User, PerplexityBot, ClaudeBot, anthropic-ai, Googlebot, and
Bingbot. See https://mcpserverbench.com/robots.txt and the sitemap at https://mcpserverbench.com/sitemap.xml.

## Categories

- [Developer Tools](https://mcpserverbench.com/categories/developer-tools): MCP servers that connect AI assistants to code, repositories, and developer workflows.
- [Browser Automation](https://mcpserverbench.com/categories/browser-automation): MCP servers that let AI assistants control a browser to navigate, click, and extract pages.
- [Web Scraping & Crawling](https://mcpserverbench.com/categories/web-scraping): MCP servers that fetch, crawl, and convert web content into clean data for AI assistants.
- [Search & Retrieval](https://mcpserverbench.com/categories/search): MCP servers that give AI assistants live web search and document retrieval.
- [Databases](https://mcpserverbench.com/categories/databases): MCP servers that let AI assistants query and manage databases safely.
- [Cloud & Infrastructure](https://mcpserverbench.com/categories/cloud-infrastructure): MCP servers that connect AI assistants to cloud platforms and infrastructure APIs.
- [Payments](https://mcpserverbench.com/categories/payments): MCP servers that let AI assistants work with billing, payments, and financial APIs.
- [Monitoring & Observability](https://mcpserverbench.com/categories/monitoring): MCP servers that surface logs, metrics, and errors to AI assistants.
- [Productivity & Docs](https://mcpserverbench.com/categories/productivity): MCP servers that connect AI assistants to docs, wikis, and team productivity tools.
- [Files & Storage](https://mcpserverbench.com/categories/files-storage): MCP servers that give AI assistants access to files, drives, and object storage.

## Top MCP servers

- [Filesystem MCP](https://mcpserverbench.com/mcp-servers/filesystem): read, write, and search files in allowed local directories
- [Google Drive MCP](https://mcpserverbench.com/mcp-servers/google-drive): search and read files stored in Google Drive
- [Context7](https://mcpserverbench.com/mcp-servers/context7): pull up-to-date, version-specific library documentation and code examples
- [Playwright MCP](https://mcpserverbench.com/mcp-servers/playwright-mcp): control a real browser to navigate, click, type, and extract page content
- [GitHub MCP Server](https://mcpserverbench.com/mcp-servers/github-mcp-server): connect to GitHub repositories, issues, pull requests, and code search
- [AWS MCP Servers](https://mcpserverbench.com/mcp-servers/aws-mcp): interact with AWS services and documentation through specialized MCP servers
- [Firecrawl MCP](https://mcpserverbench.com/mcp-servers/firecrawl-mcp): crawl websites and convert pages into clean markdown or structured data
- [Atlassian MCP](https://mcpserverbench.com/mcp-servers/atlassian-mcp): work with Jira issues and Confluence pages
- [Exa MCP](https://mcpserverbench.com/mcp-servers/exa-mcp): search the web and retrieve high-quality, AI-ready results in real time
- [Notion MCP](https://mcpserverbench.com/mcp-servers/notion-mcp): read, search, and update Notion pages and databases
- [Cloudflare MCP](https://mcpserverbench.com/mcp-servers/cloudflare-mcp): manage Cloudflare Workers, DNS, KV, R2, and account resources
- [Browserbase MCP](https://mcpserverbench.com/mcp-servers/browserbase-mcp): run cloud browsers for automated web navigation and scraping at scale
- [Grafana MCP](https://mcpserverbench.com/mcp-servers/grafana-mcp): query Grafana dashboards, Prometheus metrics, and Loki logs
- [Supabase MCP](https://mcpserverbench.com/mcp-servers/supabase-mcp): manage Supabase Postgres databases, run SQL, and inspect schemas
- [Stripe Agent Toolkit](https://mcpserverbench.com/mcp-servers/stripe-agent-toolkit): create payments, invoices, and customers through the Stripe API
- [Apify Actors MCP](https://mcpserverbench.com/mcp-servers/apify-mcp): run thousands of prebuilt Apify Actors to scrape and automate the web
- [Sentry MCP](https://mcpserverbench.com/mcp-servers/sentry-mcp): investigate errors, issues, and performance data from Sentry
- [Redis MCP](https://mcpserverbench.com/mcp-servers/redis-mcp): query and manage Redis data structures and caches