MCP Server Registry
Browse and install curated Official and Community MCP servers.
The MCP Server Registry is a curated catalog of publicly available MCP servers you can deploy on MCPLambda. It is backed by the ToolHive registry cache and groups entries into Official and Community tiers.
Use the registry when you want to discover a known server, inspect its transport and required environment variables, then install it into a project — without hunting package names or image tags yourself.
Honest positioning: The catalog is growing. Bring-your-own-server remains first-class — deploy any package (
npx/uvx/pip/go), Git repository, or container image via the deployment strategies flows. The registry is discovery and install convenience, not the only path.
What you get
- Browse, search, and filter — find servers by name, description, or tier (Official / Community).
- Server details — transport, package or image source, tools, tags, and required env vars (including which ones are secrets).
- One-click install into a project — open the create-deployment flow pre-filled from the registry entry.
In the dashboard
- Sign in to mcplambda.io.
- Open Registry in the sidebar.
- Search or filter by tier.
- Open a server to review tools, env vars, and source.
- Install into your project — the deployment form is seeded from the registry entry.
After install, the deployment behaves like any other: configure auth, secrets, profile, and connect AI clients as usual.
Public API (read, no auth)
Registry reads are public. You do not need a session or service account token to list or fetch servers.
List servers
GET https://api.mcplambda.io/v1/registry/servers
| Query param | Description |
|---|---|
search | Case-insensitive substring match on name and description |
tier | Official or Community |
tag | Exact tag match |
page | Page number (default 1) |
page_size | Results per page (default 20) |
Get one server
GET https://api.mcplambda.io/v1/registry/servers/{name}
{name} is the reverse-DNS registry name and may contain slashes (for example io.github.stacklok/fetch). Encode the path segment when calling the API.
CLI
Browse the same public catalog from the terminal with mcpl (login is still required for deploy; registry search/info use the public API).
# Search by name or description
mcpl registry search fetch
mcpl registry search github --limit 5
mcpl registry search time -o json
# Full details (transport, package/image, tools, env vars)
mcpl registry info io.github.stacklok/fetch
| Command | Description |
|---|---|
mcpl registry search <query> | Substring search; --limit defaults to 20 |
mcpl registry info <name> | Details for one entry (aliases: get, show) |
When you are ready to deploy a package or image from an entry:
mcpl deploy npx://@mcp/server-time
# or use the package/image hint printed by `mcpl registry info`
See the full mcpl CLI reference for install, login, and deploy flags.
MCPLambda MCP server
From an AI client connected to the MCPLambda MCP server, agents can search and inspect the catalog:
| Tool | Description |
|---|---|
search_registry | Search the registry; results include create_deployment mapping guidance |
get_registry_server | Fetch one registry server by name |
Typical flow: search → pick a deployable result → create secrets if required → create_deployment with the mapped fields.
Registry vs bring-your-own
| Approach | When to use |
|---|---|
| Registry install | Known community/official server; you want metadata and a pre-filled form |
| Package / Git / Image deploy | Custom server, private repo, pinned image, or anything not yet in the catalog |
All paths land on the same deployment model (URL, transport, auth, compute units).