Community

Community Registry (75+ MCPs)

The MCP Ambassador community registry ships spawn commands, environment variable schemas, and tool metadata for 75+ popular MCP servers. Admins add them to the catalog in minutes — no writing spawn commands or figuring out environment variable names from scratch.

MCP Ambassador Community Registry
Browse 75+ pre-configured MCP servers organized by category. Add any to your catalog with one click.

How to use the registry

  1. Log in to the Admin UI
  2. Navigate to MCP Catalog → Add from Registry
  3. Browse or search the registry
  4. Click Add to Catalog — spawn command and env vars are pre-filled
  5. Optionally edit the configuration (e.g., add your API key)
  6. Click Validate to verify connectivity and fetch tool list
  7. Click Publish and assign to groups

Registry contents

MCPDescriptionCredentials
tavily-searchReal-time web search optimized for AITAVILY_API_KEY
brave-searchWeb and local search via Brave Search APIBRAVE_API_KEY
exaNeural search engine for AI applicationsEXA_API_KEY

Code & Dev Tools

MCPDescriptionCredentials
githubGitHub API — search code, manage PRs, issuesGITHUB_PERSONAL_ACCESS_TOKEN
gitlabGitLab API — projects, MRs, pipelinesGITLAB_PERSONAL_ACCESS_TOKEN
sequential-thinkingDynamic, reflective problem-solvingNone
mermaidGenerate Mermaid diagrams from descriptionsNone

Data & Databases

MCPDescriptionCredentials
postgresQuery and inspect PostgreSQL databasesDATABASE_URL
sqliteRead/write SQLite database filesNone (file path)
mongodbQuery MongoDB collectionsMONGODB_URI
supabaseSupabase database and storageSUPABASE_URL, SUPABASE_KEY

Documents

MCPDescriptionCredentials
markitdownConvert any document to MarkdownNone
google-driveRead and manage Google Drive filesOAuth credentials
notionRead and write Notion pages and databasesNOTION_API_KEY

Web Automation

MCPDescriptionCredentials
fetchFetch and parse web contentNone
firecrawlCrawl and scrape websitesFIRECRAWL_API_KEY
puppeteerBrowser automation with PuppeteerNone (requires Chrome)
playwrightBrowser automation with PlaywrightNone (requires browsers)

Cloud & Infrastructure

MCPDescriptionCredentials
cloudflareManage Cloudflare zones, DNS, WorkersCLOUDFLARE_API_TOKEN
aws-kb-retrievalQuery AWS Knowledge Base (Bedrock)AWS credentials
dockerManage Docker containers and imagesDocker socket access

Business

MCPDescriptionCredentials
stripeStripe payments, customers, subscriptionsSTRIPE_API_KEY
mondayMonday.com project managementMONDAY_API_KEY
todoistTodoist task managementTODOIST_API_TOKEN
zapierTrigger Zapier automationsZAPIER_API_KEY
figmaFigma design files and componentsFIGMA_ACCESS_TOKEN
slackSend messages and manage SlackSLACK_BOT_TOKEN
linearLinear issue tracking and projectsLINEAR_API_KEY

Monitoring

MCPDescriptionCredentials
grafanaQuery Grafana dashboards and alertsGRAFANA_URL, GRAFANA_API_KEY
sentryQuery Sentry errors and eventsSENTRY_AUTH_TOKEN

Utility

MCPDescriptionCredentials
timeCurrent time and timezone conversionsNone
memoryPersistent key-value memory for agentsNone
google-mapsMaps, geocoding, directionsGOOGLE_MAPS_API_KEY
filesystemRead/write local filesystemNone (path-restricted)

Weekly sync with the official MCP Registry

The community-registry repository runs a GitHub Actions workflow, Sync official registry, that proposes new entries from the official MCP Registry.

  • Weekly dry run. Every Monday the workflow runs on a schedule. Scheduled runs are always dry runs: they build the candidate set and print what a pull request would contain, and open nothing. The first Monday of each month also walks the whole feed instead of only what changed since the last run.
  • Pull requests are manual. A maintainer starts the workflow by hand with dry run turned off to open a PR. The sync never merges, never pushes to main, and never installs or runs a candidate package. Every proposed entry is marked verified: false and must be reviewed by a person before it is merged.

Curated intake

Each run proposes a small, ranked set rather than everything in the feed:

  1. Verified publishers first. Candidates whose namespace ownership is provable (io.github.* names, or a remote whose host matches the publisher's reverse-DNS domain) are ranked ahead of the rest.
  2. Per-publisher cap. At most 2 proposed entries per publisher namespace per run by default (configurable 1-50), so no single publisher floods a batch. The overall per-run cap defaults to 10 (1-50).
  3. Repository quality score. For GitHub-hosted candidates, the sync reads the repository's public metadata. Archived repositories, and repositories with no push in the last 12 months, are rejected. The rest are scored as log10(1 + stars), plus 0.5 if the repository is at least 180 days old, and ranked best first within their tier. Candidates without a GitHub repository are unscored and rank after scored ones.

Namespace verification shows who published an entry, not that the package is safe, so reviewers still check the package, its pinned version and the permissions it needs.

Contributing to the registry

The registry is maintained in the community-registry repository.

To add an MCP to the registry:

  1. Fork the repository
  2. Add a YAML definition for your MCP in the registry/ directory
  3. Include: name, description, transport type, spawn command, environment variable definitions, and tool descriptions
  4. Submit a pull request

Registry entries are community-maintained. The maintainers review for accuracy and security before merging.

Previous
User API
Next
Contributing