What it actually does
MCP is how agents reach tools, and the ecosystem has grown faster than anyone can track informally. This is the community index of what exists.
Its practical use is negative: before you spend a week building an integration between your agent and some service, you check here and frequently find it already built. Databases, cloud providers, project trackers, communication tools, file systems, search — the coverage is broad.
It pairs with a browsable directory at glama.ai/mcp/servers if scrolling a long markdown file is not how you want to search.
92,000 stars and MIT licensed. The caveat that matters: a listing is not a security review. Every MCP server you connect gets access to whatever you grant it, and this index does not vet the code behind each entry. Read the source of anything that will touch production credentials.
Who it's for
- 01
Anyone about to build an MCP integration from scratch
- 02
Teams mapping what agent tooling is available for their stack
- 03
Engineers evaluating whether MCP covers their needs before committing
- 04
Ops leads assembling an approved internal list of vetted servers
Where it earns its keep
- Checking whether an integration exists before building it
- Discovering what agents could connect to across your existing tools
- Building an internal allowlist of reviewed, approved MCP servers
- Finding reference implementations before writing your own server
- Assessing ecosystem maturity for a tool you depend on
Use it, or skip it
Reach for it when
- You are scoping agent integrations and want to avoid rebuilding
- You need a reference implementation to model your own server on
- You are deciding whether MCP is mature enough for your stack
- You want breadth of options before picking one
Skip it when
- You need vetted, security-reviewed software — this is an index, not an audit
- Your integration is proprietary and nothing generic will fit
- You want curation over completeness; the list is long by design
- You would connect something to production without reading its source first
10 automations
Ideas, not tutorials. Each one is work a team does by hand today.
- 01Engineering
Build-versus-adopt check
Make searching this index a required step before any integration goes on the engineering backlog.
- 02Operations
Internal approved list
Review candidate servers, approve a subset, and publish that as the only list your team may install from.
- 03Operations
Stack coverage map
Map every tool your company uses against available servers to see where agent access is already possible.
- 04Engineering
New-server monitoring
Watch the repository for additions relevant to your stack and evaluate them as they appear.
- 05Engineering
Security review pipeline
Route every proposed server through a source review before it gets any production credential.
- 06Engineering
Reference implementation study
Read well-built servers in the list before writing your own internal one.
- 07Finance
Vendor pressure
Where no server exists for a tool you pay for, use that as leverage in the renewal conversation.
- 08Operations
Quarterly capability review
Re-check the index each quarter to find integrations that did not exist when you last looked.
- 09Operations
Onboarding tool inventory
Give new engineers a mapped view of what their agent can already reach in your environment.
- 10Engineering
Duplicate detection
Audit internally-built integrations against the index and retire ones a maintained server now covers.
Want one of these running by Friday?
LimeDock builds these as real workflows inside your stack — deployed to your cloud, wired into your Slack and CRM, with the code in your repo. You pay a build fee and your own API keys, nothing else.
Source
Repository stats were read from the GitHub API and reflect the last time we refreshed this entry. The editorial breakdown above is LimeDock’s own analysis — we are not affiliated with punkpeye.
https://github.com/punkpeye/awesome-mcp-servers