speclib-mcp

MCP server for searching, retrieving, and creating SpecLib specs. Allows AI agents to discover, read, and write specifications via the Model Context Protocol.

Tools

Related MCP server: AsyncAPI-MCP

Resources

Usage

One-liner

npx github:rang501/speclib-mcp

Claude Code

Add to your MCP settings (.mcp.json or project config):

{
  "mcpServers": {
    "speclib": {
      "command": "npx",
      "args": ["-y", "github:rang501/speclib-mcp"],
      "env": {
        "SPECLIB_API_URL": "http://localhost:3000",
        "SPECLIB_API_TOKEN": "your-api-token"
      }
    }
  }
}

Claude Desktop

Add to claude_desktop_config.json:

{
  "mcpServers": {
    "speclib": {
      "command": "npx",
      "args": ["-y", "github:rang501/speclib-mcp"],
      "env": {
        "SPECLIB_API_URL": "http://localhost:3000",
        "SPECLIB_API_TOKEN": "your-api-token"
      }
    }
  }
}

Local development

git clone https://github.com/rang501/speclib-mcp.git
cd speclib-mcp
npm install
node index.mjs

Configuration

Authentication

Reading public specs works without authentication. To create or update specs, you need an API token:

1. Log in to your SpecLib instance

2. Go to Account > API Tokens

3. Create a new token and copy it

4. Set it as SPECLIB_API_TOKEN in your MCP client config

Tokens that have not been used for over a year are automatically expired.

Requirements

• Node.js >= 20

• A running SpecLib instance