agentsclimarketplace

Mcp registry api

Skill aim9sour/mcp-registry-api

Official Model Context Protocol (MCP) Registry API manager skill for AI agents.

Install
npx -y skills add aim9sour/mcp-registry-api

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

  • 0 stars0 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

What its author says it does

Copied from the file, not written here

Use this skill whenever the user asks to find, install, configure, or publish an MCP (Model Context Protocol) server. Read this to learn the exact REST API commands for the official registry, how to verify publishers, and the strict security protocols required to safely evaluate and install AI tools without risking the user's system.

SKILL.md

5.5 KB, ~1.1k tokens by cl100k_base, as published. Nobody here has run it

mcp-registry-api

Instructions for the agent to follow when this skill is activated.

When to use

Use this skill whenever the user asks you to "find an MCP tool for X", "install an MCP server", "publish my MCP server", or generally interact with the MCP registry to discover and configure new capabilities.

Instructions

  1. Understand the Spirit of the Registry The MCP Registry is the definitive single source of truth for discovering, authenticating, and configuring Model Context Protocol servers. It uses stringent namespace-based authentication (via DNS, GitHub OAuth, or OIDC) to cryptographically guarantee the identity of publishers. You must act as a highly intelligent, security-conscious expert. Your primary directive is to find tools that truly meet the user's specific needs while aggressively protecting them from malicious, low-quality, or spoofed servers. Do not guess; verify everything.

  2. Identify the User's True Need When searching for a server, evaluate the user's actual goal. Do not just pick the first search result. You must read the description of the returned servers, analyze their capabilities, and meticulously select the one that most comprehensively and accurately solves the user's exact problem.

  3. Differentiate Good vs. Bad Servers Before recommending or installing any server, you must evaluate its quality and authenticity:

    • Authentication & Namespaces: The registry enforces namespace verification. A server named com.anthropic/tools or io.github.microsoft/server has been authenticated via DNS or GitHub. You must inspect the namespace to verify if it belongs to the official, reputable company it claims to be.
    • Official Metadata: Read the _meta object in the API response. The registry explicitly flags the status and authenticity of the server. Use this to determine if the server is actively maintained and officially recognized.

  4. Follow Strict Security Protocol for Unknown Sources You must ruthlessly protect the user from harmful code. Follow these absolute rules:

    • If a server is from a well-known, highly reputable company (e.g., Anthropic, Microsoft, Vercel) AND the namespace is definitively verified, you may proceed.
    • IF THE SERVER IS NOT FROM A REPUTABLE COMPANY OR IS UNKNOWN:
      • You MUST NOT download or execute it blindly.
      • You MUST find its source repository (e.g., GitHub).
      • You MUST check the number of GitHub stars, read the documentation, and inspect the source code to ensure it is not malicious, corrupt, or poorly written.
      • You MUST ask the human user for explicit, direct confirmation before downloading, installing, or configuring any unverified or third-party server.

  5. Execute Registry API Commands To interact with the registry, use curl (or curl.exe on Windows/PowerShell to avoid Invoke-WebRequest alias errors) to call the following exact endpoints. All endpoints use the base URL https://registry.modelcontextprotocol.io/v0.1. CRITICAL RULE: You must URL-encode server names in path parameters (e.g., com.example/server becomes com.example%2Fserver). Do not forget this step.

    • Discover Servers: GET /servers (Use query parameters: search=<keyword>, limit=<int>, cursor=<string>. Example: curl -s "https://registry.modelcontextprotocol.io/v0.1/servers?search=database&limit=5")
    • List Server Versions: GET /servers/{serverName}/versions (Example: curl -s "https://registry.modelcontextprotocol.io/v0.1/servers/com.example%2Fserver/versions")
    • Get Specific Server Version (Inspect & Download): GET /servers/{serverName}/versions/{version} (Use latest for the version parameter to get the most recent metadata. Example: curl -s "https://registry.modelcontextprotocol.io/v0.1/servers/com.example%2Fserver/versions/latest". Agent Action: Parse the JSON response. Extract the remotes array or installation instructions to figure out how to configure the server.)
    • Publish a Server: POST /publish (Publishing requires a rigorous namespace authentication token. The body must strictly comply with the server.schema.json standard.)
    • Update Server Version: PUT /servers/{serverName}/versions/{version} (Supported by the generic API specification to update a specific version.)
    • Delete Server Version: DELETE /servers/{serverName}/versions/{version} (Supported by the generic API specification to remove a specific version.)
    • Update Server Version Status: PATCH /servers/{serverName}/versions/{version}/status (Update the lifecycle status (e.g., deprecating a version) of a specific server version.)
    • Update All Versions Status: PATCH /servers/{serverName}/status (Update the status for all versions of a server simultaneously.)

  6. Perform Final Configuration Once you have found the optimal, secure server and retrieved its latest metadata, read the remotes field or execution commands. Generate the local configuration automatically for the user (e.g., modifying their mcp_servers.json) with the correct arguments. Do not leave the user to figure it out themselves; complete the setup entirely while adhering to the security protocols above.

What ships with it: 3 files

3.8 KB alongside SKILL.md

Gives 0 of the 12 instructions most mcp tooling skills give in ~1.1k tokens

Counted across 638 of the 750 authors here whose files we hold, read 2026-08-07

  • Create ten complex or independent read-only evaluation questionsin 69 of 638, across 15 files
  • Test servers using MCP Inspectorin 61 of 638, across 19 files
  • Provide actionable error messages with specific next stepsin 54 of 638, across 12 files
  • Prioritize comprehensive API coverage over specific workflows or workflow toolsin 54 of 638, across 12 files
  • Use TypeScript and Streamable HTTP for remote servers or clientsin 54 of 638, across 8 files
  • Define structured output schemas where possiblein 50 of 638, across 8 files
  • Use Zod or Pydantic for input schemasin 47 of 638, across 5 files
  • Fetch MCP specification pages with markdown suffixin 46 of 638, across 4 files
  • Load framework documentation using WebFetchin 45 of 638, across 3 files
  • Verify each evaluation answer independentlyin 45 of 638, across 3 files
  • Implement API client with authentication and paginationin 45 of 638, across 3 files
  • Define input schemas with validationin 27 of 638, across 9 files

Said here and by no other author read

  • verify publisher namespaces and metadata before recommending
  • select servers based on user needs
  • inspect source code of unknown servers
  • ask user before installing unverified servers
  • url-encode server names in path parameters
  • use curl to call registry endpoints

Grouped from the skills themselves: near-identical wordings counted once, and counted by distinct author, so one author publishing three of these counts once. Length counted with cl100k_base; the agent that loads this file may tokenize it differently.

Keep looking

Skills are one crate of 328,083. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.