djd-agent-score-mcp
以下摘录自该仓库的 README(片段) · 在 GitHub 读完整 README →
MCP server for **DJD Agent Score** — a reputation scoring API for AI agent wallets on Base.
This server exposes the DJD Agent Score REST API as Model Context Protocol tools, so any MCP-compatible agent (Claude, GPT, Gemini, LangChain, etc.) can call scoring endpoints natively.
Tools
| Tool | Endpoint | Cost | Description |
|------|----------|------|-------------|
| score_basic | GET /v1/score/basic | Free | Basic score, tier, confidence |
| score_full | GET /v1/score/full | $0.10 (x402) | Full dimension breakdown |
| score_refresh | GET /v1/score/refresh | $0.25 (x402) | Re-score with latest chain data |
| report_fraud | POST /v1/report | $0.02 (x402) | Submit fraud report |
| check_blacklist | GET /v1/data/fraud/blacklist | $0.05 (x402) | Check fraud reports |
| get_badge | GET /v1/badge/{wallet}.svg | Free | Embeddable SVG badge |
| get_leaderboard | GET /v1/leaderboard | Free | Top scored wallets |
| register_agent | POST /v1/agent/register | Free | Register wallet with metadata |
| health_check | GET /health | Free | System status |
Installation
Or run directly with npx:
Configuration
Claude Desktop
Add to ~/Library/Application Support/Claude/claude_desktop_config.json (macOS) or %APPDATA%\Claude\claude_desktop_config.json (Windows):
Cursor
Add to .cursor/mcp.json in your project root:
Claude Code
Add to your project's .mcp.json:
Generic MCP Client (Streamable HTTP)
Start the server in HTTP mode:
Then connect your MCP client to http://localhost:3000/mcp.
Environment Variables
| Variable | Default | Description |
|----------|---------|-------------|
| DJD_BASE_URL | https://djd-agent-score.fly.dev | API base URL (use http://localhost:3001 for local dev) |
| DJD_TIMEOUT_MS | 10000 | Request timeout in milliseconds |
| TRANSPORT | stdio | Transport mode: stdio or http |
| PORT | 3000 | HTTP server port (only used when TRANSPORT=http) |
Development
To point at a local API during development:
x402 Payment
Some endpoints require x402 micropayments. When an agent calls a paid tool, the API responds with HTTP 402 and payment instructions. Your agent framework must: