smart_toy

Agent Integration Guide

How AI agents can download resources from Mirdan.ai

descriptionllms.txtMachine-readable site info for LLMs

Mirdan.ai provides a free, public, CORS-enabled API for AI agents and tools to discover and download Claude Code resources. No authentication is required for read-only access.

For LLM crawlers: Mirdan serves /llms.txt with full API documentation. All AI user agents are explicitly allowed in robots.txt.

apiAPI Endpoints

GET/api/openclaw/skills

Search and list skills. CORS-enabled with 5-minute cache.

Query params: search, category, tag, sort, page, limit, format

GET /api/openclaw/skills?search=nextjs&limit=10
GET/api/resources

List all resource types: skills, validators, snippets.

Query params: search, type (skill|validator|block), category, tag, sort, page, limit

GET /api/resources?type=validator&sort=stars
GET/api/resources/{name}

Get a single resource by its slug name. Returns full content in the readme field.

GET /api/resources/test-driven-development
GET/api/tools

List MCP tools with install commands and configuration.

Query params: search, category, featured, page, limit

GET /api/tools?category=database
POST/api/contribute

Submit new resources. Requires auth via session cookie or x-api-key header.

POST /api/contribute
Headers: x-api-key: YOUR_API_KEY
Body: { type, name, displayName, description, readme, tags, categories }

downloadHow to Install Each Resource Type

codeSkills

  1. Fetch the skill: GET /api/resources/{name}
  2. Extract the readme field (contains the full skill.md content)
  3. Save to .claude/skills/{name}.md in the project directory
  4. Or save to ~/.claude/skills/{name}.md for global use across all projects
# Example with curl
curl -s https://mirdan.ai/api/resources/test-driven-development \
  | jq -r '.data.readme' > .claude/skills/test-driven-development.md

verified_userValidators

  1. Fetch: GET /api/resources/{name}
  2. Extract the readme field (contains the validator.md content)
  3. Save to .claude/skills/{name}.md (validators use the same skills directory)
curl -s https://mirdan.ai/api/resources/code-reviewer \
  | jq -r '.data.readme' > .claude/skills/code-reviewer.md

articleSnippets

  1. Fetch: GET /api/resources/{name}
  2. Check the categories array for the target file: "claude.md", "memory.md", or "settings.md"
  3. Append the readme content to the appropriate config file:

claude.md → Project root (team-shared via repo)

memory.md → ~/.claude/memory.md (personal, persistent)

settings.md → ~/.claude/settings.md (global, all projects)

buildMCP Tools

  1. Fetch tool config: GET /api/tools/{name}
  2. Extract command, args, and env fields
  3. Add to Claude Code's settings.json under the mcpServers key
// Add to ~/.claude/settings.json
{
  "mcpServers": {
    "github": {
      "command": "npx",
      "args": ["@modelcontextprotocol/server-github"]
    }
  }
}

data_objectResponse Format

{
  "success": true,
  "data": {
    "id": "uuid",
    "name": "test-driven-development",
    "displayName": "Test-Driven Development",
    "type": "skill",
    "description": "Enforce strict TDD workflow...",
    "readme": "---\nname: test-driven-development\n...full content...",
    "version": "1.0.0",
    "license": "MIT",
    "tags": ["testing", "tdd", "workflow"],
    "categories": ["Testing"],
    "author": {
      "username": "mirdan",
      "displayName": "Mirdan"
    },
    "stats": {
      "stars": 42,
      "downloads": 1200,
      "forks": 5
    }
  }
}

descriptionllms.txt

Mirdan serves an /llms.txt file following the emerging standard for LLM-readable site information. It contains:

  • checkFull API endpoint documentation
  • checkResource type descriptions and installation instructions
  • checkQuery parameter reference for each endpoint
  • checkKey site URLs for navigation

Build with the Mirdan API

Integrate Mirdan's resource catalog into your AI agent, CLI tool, or IDE extension. Free and open for all.

exploreBrowse Resourcesdescription/llms.txt