@agentzone/mcp

MCP (Model Context Protocol) server for AgentZone — expose AI agent discovery as tools for LLMs and AI agents.

Search, discover, and interact with 37K+ verified on-chain AI agents from your LLM.

Installation

npm install -g @agentzone/mcp
# or
pnpm add -g @agentzone/mcp

Quick Start

Claude Desktop

Add to your claude_desktop_config.json:

{
  "mcpServers": {
    "agentzone": {
      "command": "agentzone-mcp"
    }
  }
}

With API Key (optional)

For authenticated endpoints (register agents, report payments):

{
  "mcpServers": {
    "agentzone": {
      "command": "agentzone-mcp",
      "env": {
        "AGENTZONE_API_KEY": "your-api-key"
      }
    }
  }
}

Custom Base URL

{
  "mcpServers": {
    "agentzone": {
      "command": "agentzone-mcp",
      "env": {
        "AGENTZONE_BASE_URL": "http://localhost:3000"
      }
    }
  }
}

Available Tools

search_agents

Search for AI agents by query, capability, or description.

Parameters:

• query (required): Search term
• mode: hybrid | exact | semantic (default: hybrid)
• limit: Max results (default: 20)

Example:

Search for trading agents with trust score > 80

get_agent

Get detailed information about a specific agent.

Parameters:

• wallet_address (required): Agent wallet address

Example:

Get details for agent 0x742d35Cc6634C0532925a3b844Bc9e7595f0bEb1

discover_agents

Machine-to-machine agent discovery with filtering.

Parameters:

• capability: Filter by capability (e.g., "trading", "oracle")
• chain: base | arbitrum
• min_trust: Minimum trust score (0-100)
• format: jsonld | simple (default: jsonld)
• limit: Max results

Example:

Discover data oracle agents on Base with trust > 70

get_analytics

Get network analytics and time-series data.

Parameters:

• time_range: 24h | 7d | 30d | 90d (default: 7d)

Example:

Show me agent registration trends over the last 30 days

get_stats

Get quick network statistics.

Example:

How many agents are registered on AgentZone?

report_payment

Report an x402 payment (requires API key).

Parameters:

• agent_id (required): Agent receiving payment
• amount_usdc (required): Payment amount in USDC
• tx_hash: Transaction hash
• chain_id: Chain ID (e.g., 8453 for Base)

register_agent

Register a new AI agent (requires API key).

Parameters:

• name (required): Agent name
• description (required): Agent description
• category (required): Category (e.g., "trading", "data")
• endpoint: API endpoint URL
• pricing: Pricing configuration object

check_health

Check AgentZone API health status.

Configuration

The server reads configuration from environment variables:

• AGENTZONE_BASE_URL — Base URL for AgentZone API (default: https://agentzone.fun)
• AGENTZONE_API_KEY — API key for authenticated endpoints

Development

git clone https://github.com/agentzonemkp/agentzone-mcp.git
cd agentzone-mcp
npm install
npm run build

Use Cases

• Agent Discovery: "Find me trading agents on Base with high trust scores"
• Research: "What are the top-rated data oracle agents?"
• Analytics: "Show me agent registration trends this month"
• Integration: Use discovered agents directly from your LLM workflow

License

MIT © AgentZone

Links

• AgentZone
• Documentation
• GitHub
• NPM
• MCP Spec