Best MCP registry

Combined ranking

1. 1
   Official MCP Registry↗—15 pts

   GPT #3Claude #2Gemini #3Grok #1

   WHY: Canonical source of truth with live handshake validation, broad ecosystem backing (e.g., Anthropic, GitHub contributors), active recent updates, open API for sub-registries, and direct discoverability for public MCP servers—highest legitimacy and coverage for typical practitioners seeking reliable tools without vendor lock-in. FIX: Preview-stage with potential instability/incompleteness; less emphasis on enterprise governance or private curation.

   + model takes & fixes− hide details

   Grok WHY: Canonical source of truth with live handshake validation, broad ecosystem backing (e.g., Anthropic, GitHub contributors), active recent updates, open API for sub-registries, and direct discoverability for public MCP servers—highest legitimacy and coverage for typical practitioners seeking reliable tools without vendor lock-in. FIX: Preview-stage with potential instability/incompleteness; less emphasis on enterprise governance or private curation.

   Claude The canonical, open-source, API-first source of truth backed by the MCP steering committee (Anthropic, GitHub, et al.), with namespace-verified publishers — the upstream that most other directories now consume, so it has the best provenance and least SEO junk.

   GPT The strongest source of canonical publisher metadata, verified namespace ownership, versioned distributions, an open API, and ecosystem-neutral infrastructure; the best starting point when authenticity and machine-readable discovery matter most.

   Gemini Backed by the protocol's steering group as the canonical, DNS-verified source of truth for public metadata schemas, ensuring namespace authority and registry persistence.

   Where it falls short

   per GPT It is a registry backbone rather than a mature recommendation marketplace, with weaker ranking, evaluation, testing, and install UX than Glama or Smithery.

   per Claude It is a metadata backbone, not a shopping experience — minimal browsing UX, no hosting, ratings, or curation, so most practitioners consume it through a downstream client or aggregator rather than directly.

   per Gemini It lacks a user-friendly GUI, categorized discovery tools, or developer sandboxes, serving strictly as a raw metadata API repository.

   per Grok Preview-stage with potential instability/incompleteness; less emphasis on enterprise governance or private curation.

2. 2
   Smithery↗—9 pts

   GPT #2Claude #1Gemini —Grok —

   The most complete practitioner marketplace — thousands of indexed MCP servers with usage stats, a CLI that installs/configures servers into Claude, Cursor, and other clients in one command, and hosted remote servers so you can try a tool without running it locally; assumption: the typical practitioner wants discover-to-running-in-minutes, which Smithery does best end to end.

   + model takes & fixes− hide details

   Claude The most complete practitioner marketplace — thousands of indexed MCP servers with usage stats, a CLI that installs/configures servers into Claude, Cursor, and other clients in one command, and hosted remote servers so you can try a tool without running it locally; assumption: the typical practitioner wants discover-to-running-in-minutes, which Smithery does best end to end.

   GPT Near-tied with Glama; combines a substantial catalog with excellent installation, configuration, remote hosting, and deployment ergonomics, making discovery unusually likely to end in a working integration.

   Where it falls short

   per GPT Its managed-platform emphasis and limited independent security assurance are poor fits for teams requiring rigorous vetting or fully self-controlled infrastructure.

   per Claude Quality and trust vary widely across its long tail, and running third-party hosted servers means routing your data/credentials through infrastructure you don't control — security-sensitive teams should vet or self-host instead.

3. 3
   Glama↗—5 pts

   GPT #1Claude —Gemini —Grok —

   Best overall discovery workflow: exceptionally broad coverage, tool-level indexing, quality and safety signals, browser testing, install guidance, and optional hosted deployment; narrowly beats Smithery for practitioners comparing unfamiliar tools.

   + model takes & fixes− hide details

   GPT Best overall discovery workflow: exceptionally broad coverage, tool-level indexing, quality and safety signals, browser testing, install guidance, and optional hosted deployment; narrowly beats Smithery for practitioners comparing unfamiliar tools.

   Where it falls short

   per GPT Automated scoring and scanning reduce risk but cannot establish that community servers are trustworthy or well designed.

4. 4
   Glama.ai↗—5 pts

   GPT —Claude —Gemini #1Grok —

   Near-tie with Smithery.ai; Glama.ai takes the top spot due to its built-in interactive in-browser testing inspector, automated security/compatibility grading, and developer API gateway, which solve the critical developer trust deficit.

   + model takes & fixes− hide details

   Gemini Near-tie with Smithery.ai; Glama.ai takes the top spot due to its built-in interactive in-browser testing inspector, automated security/compatibility grading, and developer API gateway, which solve the critical developer trust deficit.

   Where it falls short

   per Gemini The automated scoring system can create false confidence about security, and it lacks automated local client configuration tools.

5. 5
   PulseMCP↗—4 pts

   GPT #4Claude #4Gemini —Grok —

   A large, frequently updated directory with useful search, topical discovery, remote-server filtering, and distinctions among official-provider, reference, and community entries.

   + model takes & fixes− hide details

   GPT A large, frequently updated directory with useful search, topical discovery, remote-server filtering, and distinctions among official-provider, reference, and community entries.

   Claude The best-curated pure directory — rich per-server metadata, dedup of forks/clones, download and popularity trend data, and an editorial newsletter that surfaces genuinely useful new servers; near-tie with GitHub's registry, ranked ahead on breadth and metadata depth.

   Where it falls short

   per GPT Breadth exceeds curation: listings generally lack the testing, deployment workflow, and substantive trust evidence needed to choose safely on their own.

   per Claude Discovery only — no install tooling, hosting, or security vetting, so everything after "found it" is still on you.

6. 6
   Aagentregistry—4 pts

   GPT —Claude —Gemini —Grok #2

   WHY: Open-source, developer-centric with seamless CLI/UI deploy, IDE integration (Claude, Cursor, VS Code), curation of agents/skills/prompts alongside MCP servers, gateway for secure access, and local-to-prod workflow—strongest real-world value for practitioners building and sharing internal/team artifacts. FIX: Newer platform; requires setup of its daemon/gateway for full benefits, not purely a passive public directory.

   + model takes & fixes− hide details

   Grok WHY: Open-source, developer-centric with seamless CLI/UI deploy, IDE integration (Claude, Cursor, VS Code), curation of agents/skills/prompts alongside MCP servers, gateway for secure access, and local-to-prod workflow—strongest real-world value for practitioners building and sharing internal/team artifacts. FIX: Newer platform; requires setup of its daemon/gateway for full benefits, not purely a passive public directory.

   Where it falls short

   per Grok Newer platform; requires setup of its daemon/gateway for full benefits, not purely a passive public directory.

7. 7
   SSmithery.ai—4 pts

   GPT —Claude —Gemini #2Grok —

   Near-tie with Glama.ai; Smithery.ai earns its spot by prioritizing developer utility with its CLI-first installation workflow (@smithery/cli) for local clients and offering hosted remote server execution with tool telemetry.

   + model takes & fixes− hide details

   Gemini Near-tie with Glama.ai; Smithery.ai earns its spot by prioritizing developer utility with its CLI-first installation workflow (@smithery/cli) for local clients and offering hosted remote server execution with tool telemetry.

   Where it falls short

   per Gemini It abstracts configuration so heavily that troubleshooting is difficult when the automated installer fails to merge configs correctly.

8. 8
   Docker MCP Catalog↗—3 pts

   GPT —Claude #3Gemini —Grok —

   Containerized, signature-verified, security-scanned MCP servers distributed through Docker Hub with the MCP Toolkit handling sandboxed local execution and secrets — the strongest answer to the supply-chain-risk problem that plagues every other registry, and the default pick for enterprise/security-conscious teams.

   + model takes & fixes− hide details

   Claude Containerized, signature-verified, security-scanned MCP servers distributed through Docker Hub with the MCP Toolkit handling sandboxed local execution and secrets — the strongest answer to the supply-chain-risk problem that plagues every other registry, and the default pick for enterprise/security-conscious teams.

   Where it falls short

   per Claude Covers only the curated Dockerized subset of the ecosystem and assumes a Docker Desktop/daemon workflow, so it lags new servers and is overkill for quick experiments.

9. 9
   KKong MCP Registry—3 pts

   GPT —Claude —Gemini —Grok #3

   WHY: Enterprise-grade governance, dynamic discovery, observability, and integration with existing API/security stacks to prevent shadow AI—excellent control and visibility for teams needing trusted, approved tool access at scale. FIX: Commercial/enterprise-focused, potentially overkill or costly for solo/open-source practitioners.

   + model takes & fixes− hide details

   Grok WHY: Enterprise-grade governance, dynamic discovery, observability, and integration with existing API/security stacks to prevent shadow AI—excellent control and visibility for teams needing trusted, approved tool access at scale. FIX: Commercial/enterprise-focused, potentially overkill or costly for solo/open-source practitioners.

   Where it falls short

   per Grok Commercial/enterprise-focused, potentially overkill or costly for solo/open-source practitioners.

10. 10
    JJFrog MCP Registry—2 pts

    GPT —Claude —Gemini —Grok #4

    WHY: Unified with proven software supply chain security (scanning, policies, AI Catalog integration), strong versioning/governance for vetted servers—valuable for security-conscious orgs treating MCP as part of the artifact lifecycle. FIX: Heavily enterprise-oriented with focus on binaries/supply chain; less agile for rapid public discovery or lightweight dev use.

    + model takes & fixes− hide details

    Grok WHY: Unified with proven software supply chain security (scanning, policies, AI Catalog integration), strong versioning/governance for vetted servers—valuable for security-conscious orgs treating MCP as part of the artifact lifecycle. FIX: Heavily enterprise-oriented with focus on binaries/supply chain; less agile for rapid public discovery or lightweight dev use.

    Where it falls short

    per Grok Heavily enterprise-oriented with focus on binaries/supply chain; less agile for rapid public discovery or lightweight dev use.

11. 11
    MMCP.so—2 pts

    GPT —Claude —Gemini #4Grok —

    Provides a highly readable, consumer-friendly web catalog indexing a massive volume of community servers with clean categorizations and instant copy-paste configuration snippets.

    + model takes & fixes− hide details

    Gemini Provides a highly readable, consumer-friendly web catalog indexing a massive volume of community servers with clean categorizations and instant copy-paste configuration snippets.

    Where it falls short

    per Gemini It lacks active security scanning, automated validation pipelines, or interactive runtime environments.

12. 12
    GGitHub MCP Registry—1 pts

    GPT —Claude #5Gemini —Grok —

    A curated, high-trust catalog tied to real repos with one-click install into VS Code/Copilot, and self-service publishing wired to the official registry — quality bar per listing is among the highest anywhere.

    + model takes & fixes− hide details

    Claude A curated, high-trust catalog tied to real repos with one-click install into VS Code/Copilot, and self-service publishing wired to the official registry — quality bar per listing is among the highest anywhere.

    Where it falls short

    per Claude Deliberately small and GitHub/Copilot-centric; if your client isn't in that ecosystem or you need long-tail coverage, it won't have what you're looking for.

13. 13
    Mmcp.so—1 pts

    GPT #5Claude —Gemini —Grok —

    Very broad coverage, approachable browsing, tags, and copyable client configuration make it useful for finding obscure or long-tail MCP servers that smaller curated catalogs miss.

    + model takes & fixes− hide details

    GPT Very broad coverage, approachable browsing, tags, and copyable client configuration make it useful for finding obscure or long-tail MCP servers that smaller curated catalogs miss.

    Where it falls short

    per GPT Inconsistent listing quality, duplication, and limited security or maintenance verification make it unsuitable as the sole basis for installing privileged tools.

14. 14
    MMCP.SO—1 pts

    GPT —Claude —Gemini —Grok #5

    WHY: Community-driven marketplace with usage rankings, filters, and broad curated directory—practical for quick browsing and trending real-world adoption insights. FIX: Popularity/traffic bias over verified quality or governance; more aggregator than authoritative source.

    + model takes & fixes− hide details

    Grok WHY: Community-driven marketplace with usage rankings, filters, and broad curated directory—practical for quick browsing and trending real-world adoption insights. FIX: Popularity/traffic bias over verified quality or governance; more aggregator than authoritative source.

    Where it falls short

    per Grok Popularity/traffic bias over verified quality or governance; more aggregator than authoritative source.

15. 15
    Ppunkpeye/awesome-mcp-servers—1 pts

    GPT —Claude —Gemini #5Grok —

    Offers high peer-review standards as a community-maintained GitHub repository, showcasing production-tested open-source implementations without platform monetization or tracking.

    + model takes & fixes− hide details

    Gemini Offers high peer-review standards as a community-maintained GitHub repository, showcasing production-tested open-source implementations without platform monetization or tracking.

    Where it falls short

    per Gemini It is a static Markdown file that lacks programmatic search APIs, status checks, and interactive playgrounds.

Just missed the top 5

By model

Tracked by ModelsAgree · rank 1 = 5 pts … rank 5 = 1 pt · re-polled weekly