token-rugcheck
MCP server for real-time Solana token risk analysis. Cross-references RugCheck.xyz, DexScreener, and GoPlus Security to generate three-layer reports: machine verdict → LLM analysis → raw on-chain evidence. Live on Solana mainnet with USDC micropayments ($0.02/audit). Give any AI agent the ability to check if a token is safe before trading.
README
Token RugCheck — Solana Token Safety Audit for AI Agents
We're LIVE on Mainnet! Try it now:
https://rugcheck.aethercore.dev
# Quick test — no setup needed
curl https://rugcheck.aethercore.dev/health
# See the 402 paywall in action
curl https://rugcheck.aethercore.dev/v1/audit/DezXAZ8z7PnrnRJjz3wXBoRgixCa6xjnB7YaB1pPB263
What you get: A three-layer safety audit for any Solana token — machine-readable verdict, LLM-friendly analysis, and raw evidence — all in one API call for $0.02 USDC.
Powered by ag402 on-chain micropayments.
Disclaimer: Not financial advice (NFA). Token safety scores are automated heuristics — may be inaccurate or outdated. DYOR. The authors accept no liability for losses.
Live Demos (Mainnet)
All demos below are real mainnet transactions — left side is the client, right side is the production server logs.
1. Wallet Setup (ag402 setup)
https://github.com/user-attachments/assets/fb6b1ecb-8d42-43b7-9939-4751ca09b63f
2. Auditing a Safe Token (BONK — risk score 3, SAFE)
https://github.com/user-attachments/assets/4b3814ae-96af-496b-86b7-c80cddef1475
3. Auditing a Risky Token (TRUMP — risk score 60, HIGH)
https://github.com/user-attachments/assets/6e359374-8caf-41c7-8bd1-14f63eb6d6e8
How It Works
Your AI Agent RugCheck Service
│ │
│ GET /v1/audit/{mint} │
├─────────────────────────────────────▶│
│ │
│ 402 Payment Required │
│ (pay 0.02 USDC on Solana) │
│◀─────────────────────────────────────┤
│ │
│ USDC payment (on-chain) │
├─────────────────────────────────────▶│
│ │
│ 200 OK + Audit Report │
│◀─────────────────────────────────────┤
Input a Solana token mint address → get a three-layer report:
| Layer | For | Content |
|---|---|---|
| Action | Machines | is_safe, risk_score (0-100), risk_level (SAFE/LOW/MEDIUM/HIGH/CRITICAL) |
| Analysis | LLMs | Summary, red flags, green flags |
| Evidence | Humans | Price, liquidity, holder distribution, mint/freeze authority, raw data |
Data sources: RugCheck.xyz + DexScreener + GoPlus Security (concurrent fetch, graceful degradation).
Try It Now (3 minutes)
Step 1: Install
pip install "ag402-core[crypto]" httpx
Step 2: Set up your wallet
ag402 setup
# Choose: Consumer → Mainnet
# Enter your Solana private key (encrypted locally with AES)
# Set safety limits (default: $10/day max)
Step 3: Run an audit
# Python one-liner
python3 -c "
import asyncio, httpx, ag402_core
ag402_core.enable()
async def run():
async with httpx.AsyncClient(timeout=30.0) as c:
r = await c.get('https://rugcheck.aethercore.dev/v1/audit/DezXAZ8z7PnrnRJjz3wXBoRgixCa6xjnB7YaB1pPB263')
print(r.json()['action'])
asyncio.run(run())
"
Or in Python:
import asyncio, httpx, ag402_core
ag402_core.enable() # Auto-handles 402 → pay → retry
async def check_token(mint: str):
async with httpx.AsyncClient(timeout=30.0) as client:
resp = await client.get(
f"https://rugcheck.aethercore.dev/v1/audit/{mint}"
)
report = resp.json()
action = report["action"]
if not action["is_safe"]:
print(f"DANGER — risk score {action['risk_score']}/100")
for flag in report["analysis"]["red_flags"]:
print(f" 🚩 {flag['message']}")
else:
print(f"SAFE — {report['analysis']['summary']}")
asyncio.run(check_token("DezXAZ8z7PnrnRJjz3wXBoRgixCa6xjnB7YaB1pPB263"))
Requirements: Your wallet needs USDC (for payments) and a small amount of SOL (for transaction fees, ~0.01 SOL).
Command-line Audit Tool (ag402 pay)
You can audit any token directly from the command line using the ag402 CLI:
# Single request with auto-payment
ag402 pay https://rugcheck.aethercore.dev/v1/audit/DezXAZ8z7PnrnRJjz3wXBoRgixCa6xjnB7YaB1pPB263
# Or use the Python snippet above with a custom gateway
Note:
mainnet_buyer_test.py(referenced in OPERATIONS.md history) is a local wallet test script intentionally excluded from the repo via.gitignore— it contains wallet-specific configuration not suitable for distribution.
API Reference
Base URL: https://rugcheck.aethercore.dev
Endpoints
| Method | Path | Auth | Description |
|---|---|---|---|
| GET | /v1/audit/{mint_address} |
USDC payment | Full safety audit report |
| GET | /health |
None | Service health + upstream status |
| GET | /stats |
Loopback only | Request counts + cache hit rate |
| GET | /metrics |
Loopback only | Prometheus metrics |
| GET | /audit/{mint_address} |
USDC payment | Deprecated — use /v1/ |
Audit Response Schema
{
"contract_address": "DezXAZ8z7PnrnRJjz3wXBoRgixCa6xjnB7YaB1pPB263",
"chain": "solana",
"audit_timestamp": "2026-03-08T12:34:56.789012+00:00",
"degraded": false,
"action": {
"is_safe": true,
"risk_level": "SAFE",
"risk_score": 3
},
"analysis": {
"summary": "No significant risk signals detected. Always do your own research (DYOR).",
"red_flags": [
{"level": "LOW", "message": "Token metadata is mutable — common for Solana tokens."}
],
"green_flags": [
{"message": "Mint authority renounced (Mint Renounced)."},
{"message": "Liquidity pool is sufficiently protected (LP Burned or Locked)."},
{"message": "No freeze authority (Not Freezable)."}
]
},
"evidence": {
"token_name": "Bonk",
"token_symbol": "Bonk",
"price_usd": 0.00012,
"liquidity_usd": 85000000.0,
"is_mintable": false
},
"metadata": {
"data_sources": ["RugCheck", "DexScreener", "GoPlus"],
"data_completeness": "full",
"cache_hit": false,
"data_age_seconds": 0,
"response_time_ms": 738,
"disclaimer": "This report is generated by automated data aggregation. Not financial advice (NFA)."
}
}
Key Fields
| Field | Type | Description |
|---|---|---|
action.is_safe |
bool |
Machine-readable verdict |
action.risk_score |
int |
0 (safest) to 100 (most dangerous) |
action.risk_level |
string |
SAFE / LOW / MEDIUM / HIGH / CRITICAL |
degraded |
bool |
true if upstream sources failed — treat with extra caution |
metadata.data_completeness |
string |
full / partial / minimal / unavailable |
metadata.data_age_seconds |
int|null |
0 = fresh, >0 = cached |
Health Check
# Development mode — detailed info
GET /health → {"status": "ok", "service": "token-rugcheck-mcp", "version": "0.1.0"}
# Production mode (RUGCHECK_PRODUCTION=true) — minimal, no internal details
GET /health → {"status": "ok"}
status |
Meaning |
|---|---|
ok |
All systems normal |
degraded |
Upstream API failures — service continues with available data |
Deployment Guide (Self-hosting)
Automated Deployment: See OPERATIONS.md for one-click scripts and ops runbook.
Architecture
Client (AI Agent)
│ HTTPS
▼
Cloudflare (SSL termination, DDoS protection)
│ HTTP:80
▼
┌─────────────────────────────────────────────┐
│ Docker Compose │
│ ┌─────────────┐ ┌──────────────────┐ │
│ │ ag402-gateway│────▶│ rugcheck-audit │ │
│ │ :80 (public) │ │ :8000 (internal) │ │
│ │ Payment gate │ │ Audit engine │ │
│ └─────────────┘ └──────────────────┘ │
└─────────────────────────────────────────────┘
Three Environments
| Environment | Blockchain | Real Funds | Use Case |
|---|---|---|---|
| Test (mock) | None | No | Local dev, CI |
| Devnet | Solana Devnet | No (faucet) | Integration testing |
| Production | Solana Mainnet | Yes | Live service |
Quick Start — Test Mode (zero config)
pip install -e .
python -m rugcheck.main & # Audit server on :8000
python -m rugcheck.gateway & # Gateway on :8001 (mock payments)
curl http://localhost:8001/v1/audit/DezXAZ8z7PnrnRJjz3wXBoRgixCa6xjnB7YaB1pPB263
Quick Start — Production (Docker)
# 1. One-click deploy to your server
bash scripts/deploy-oneclick.sh
# Or manually:
cp .env.example .env # Edit with your wallet address + keys
docker compose -f docker-compose.yml -f docker-compose.prod.yml up -d
Provider .env (Production)
X402_MODE=production
X402_NETWORK=mainnet
AG402_ADDRESS=<your_solana_wallet> # Receives USDC payments
AG402_PRICE=0.02 # USDC per audit
SOLANA_RPC_URL=https://api.mainnet-beta.solana.com # Read-only RPC for payment verification
RUGCHECK_PRODUCTION=true # Disable /docs, harden /health response
UVLOOP_INSTALL=0 # Required: prevents uvloop/aiosqlite crash
# Optional: prepaid fast-path (~1ms per request instead of ~500ms)
# Generate: python -c "import secrets; print(secrets.token_hex(32))"
AG402_PREPAID_SIGNING_KEY=<random_32+_char_key>
# Optional: higher GoPlus rate limits
GOPLUS_APP_KEY=<key>
GOPLUS_APP_SECRET=<secret>
Note: As a seller/provider, you do not need
SOLANA_PRIVATE_KEY. The gateway verifies buyer payments via read-only RPC — no signing required.
Consumer Setup
pip install "ag402-core[crypto]" httpx
ag402 setup # Interactive wizard
ag402 status # Verify wallet + balance
import ag402_core
ag402_core.enable()
# Now all httpx requests auto-handle 402 → pay → retry
Configuration Reference
Service
| Variable | Default | Description |
|---|---|---|
RUGCHECK_HOST |
0.0.0.0 |
Bind address |
RUGCHECK_PORT |
8000 |
Bind port |
RUGCHECK_LOG_LEVEL |
info |
debug / info / warning / error |
RUGCHECK_PRODUCTION |
false |
true disables /docs, /redoc, /openapi.json |
UVICORN_WORKERS |
1 |
Worker processes |
CACHE_TTL_SECONDS |
3 |
Cache TTL (short to catch rug pulls) |
CACHE_MAX_SIZE |
5000 |
Max cached entries (LRU) |
FREE_DAILY_QUOTA |
20 |
Free requests per IP per day |
PAID_RATE_LIMIT |
120 |
Paid requests per IP per minute |
Upstream APIs
| Variable | Default | Description |
|---|---|---|
DEXSCREENER_TIMEOUT_SECONDS |
1.5 |
DexScreener timeout |
GOPLUS_TIMEOUT_SECONDS |
2.5 |
GoPlus timeout |
RUGCHECK_API_TIMEOUT_SECONDS |
3.5 |
RugCheck timeout |
GOPLUS_APP_KEY |
— | GoPlus API key (optional) |
GOPLUS_APP_SECRET |
— | GoPlus API secret |
ag402 Payment
| Variable | Default | Description |
|---|---|---|
AG402_PRICE |
0.02 |
USDC per request |
AG402_ADDRESS |
— | Provider wallet (receives payments) |
AG402_GATEWAY_PORT |
8001 |
Gateway port |
AG402_PREPAID_SIGNING_KEY |
— | HMAC signing key for prepaid fast-path (optional, >=32 chars) |
X402_MODE |
test |
test (mock) / production (real) |
X402_NETWORK |
devnet |
mock / devnet / mainnet |
SOLANA_RPC_URL |
https://api.devnet.solana.com |
Read-only Solana RPC for payment verification |
Consumer Safety Limits
| Variable | Default | Description |
|---|---|---|
X402_DAILY_LIMIT |
10.0 |
Max daily spend (USD) |
X402_SINGLE_TX_LIMIT |
5.0 |
Max per-transaction |
X402_PER_MINUTE_LIMIT |
2.0 |
Max spend per minute |
X402_PER_MINUTE_COUNT |
5 |
Max transactions per minute |
ag402 CLI Reference
ag402 setup # Interactive setup wizard
ag402 status # Dashboard: mode, wallet, balance
ag402 balance # Check SOL + USDC balance
ag402 doctor # Diagnose environment issues
ag402 history --limit 10 # Recent transactions
ag402 pay <url> # Send a single paid request
ag402 demo # Quick E2E test (mock mode)
ag402 demo --devnet # E2E test with Devnet transactions
ag402 info # Protocol version
# Prepaid packages (v0.1.15+) — pre-purchase call bundles for ~1ms per request
ag402 prepaid buy <gateway_url> <package_id> # Purchase a prepaid package
ag402 prepaid status # List all credentials + remaining calls
ag402 prepaid purge # Remove expired/depleted credentials
ag402 prepaid pending # Show in-flight purchase (if any)
ag402 prepaid recover <gateway_url> # Recover credential after timeout/network failure
Development
pip install -e ".[dev]"
python -m pytest tests/ -v # 119 tests
ruff check src/ tests/ # Lint
python examples/demo_agent.py # E2E demo (direct)
python examples/demo_agent.py --with-gateway # E2E demo (with payment)
Project Structure
src/rugcheck/
├── config.py # Environment-based configuration
├── models.py # Pydantic models (report schema)
├── cache.py # Async-safe TTL cache (LRU, asyncio.Lock)
├── server.py # FastAPI app + rate limiter + health checks
├── main.py # Audit server entry point
├── gateway.py # ag402 gateway entry point
├── fetchers/
│ ├── base.py # BaseFetcher ABC
│ ├── goplus.py # GoPlus Security API
│ ├── rugcheck.py # RugCheck.xyz API
│ ├── dexscreener.py # DexScreener API
│ └── aggregator.py # Concurrent fetch + merge
└── engine/
└── risk_engine.py # Deterministic rule-based scoring
Security
- Rate limiting — free: 20/day per IP; paid: 120/min per IP
- Trusted proxy model —
CF-Connecting-IPonly trusted from Cloudflare IPs - Production hardening —
/docs,/redoc,/openapi.jsondisabled - Gateway fail-safe — refuses to start if payment verifier fails in production
- Cache isolation — deep-copy on get/set prevents shared state corruption
- Degraded short-TTL — incomplete reports cached only 10s
- Prometheus path normalization — prevents cardinality explosion
- Upstream protection —
Semaphore(20)+max_connections=50 - Error sanitization — never exposes internal paths
Troubleshooting
| Problem | Cause | Fix |
|---|---|---|
402 Payment Required |
Expected — gateway requires USDC payment | Use ag402_core.enable() or ag402 pay <url> |
InsufficientFundsForRent |
Wallet SOL too low for ATA creation | Send ≥ 0.01 SOL to your wallet |
Payment not confirmed on-chain (403) |
Solana network confirmation timeout | Retry — transient network issue |
ValueError: Production mode requires PaymentVerifier |
Missing crypto deps | pip install -e ".[crypto]" |
ImportError: solana/solders |
Missing crypto deps | pip install "ag402-core[crypto]" |
| Gateway keeps restarting | Missing SOLANA_PRIVATE_KEY or AG402_ADDRESS |
Check .env configuration |
Health returns degraded |
Upstream APIs failing | Service continues with available sources — check /health |
License
MIT
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。