Skip to main content
Glama
README.md
# RugRadar MCP

Solana token rug and scam risk checks for AI agents, over the [Model Context Protocol](https://modelcontextprotocol.io).

Give your agent a read-only check before it touches an unfamiliar Solana token: a 0-100 risk score (0-29 low risk, 30-59 caution, 60-100 high risk) with plain-English red flags, from on-chain data, RugCheck, Jupiter and DexScreener. It never needs a wallet connection, signature, private key or seed phrase.

**Automated risk signals, not financial advice.** A low score means no major red flags were found, not that a token is safe.

## Hosted endpoint (recommended)

```
https://rugradar-bgcm.onrender.com/mcp
```

Streamable HTTP, stateless, no auth for the free tier. Pro key: `Authorization: Bearer <key>`.

**Claude Code**

```bash
claude mcp add --transport http rugradar https://rugradar-bgcm.onrender.com/mcp
```

**Cursor** (`~/.cursor/mcp.json`)

```json
{ "mcpServers": { "rugradar": { "url": "https://rugradar-bgcm.onrender.com/mcp" } } }
```

**Claude Desktop**: Settings → Connectors → Add custom connector → the URL above. Or use this package over stdio (below).

## This package (stdio bridge)

A ~50-line stdio server that forwards every tool call to the hosted endpoint. Use it with clients that only speak stdio. No build step, no secrets, nothing stored.

```json
{
  "mcpServers": {
    "rugradar": {
      "command": "npx",
      "args": ["-y", "github:HOBSTER6615/rugradar-mcp"],
      "env": { "RUGRADAR_API_KEY": "" }
    }
  }
}
```

| Env | Meaning |
| --- | --- |
| `RUGRADAR_API_KEY` | Optional Pro key, sent as `Authorization: Bearer` |
| `RUGRADAR_MCP_URL` | Optional endpoint override |

## Tools

| Tool | What it does |
| --- | --- |
| `check_token` | Risk score, red flags (mint/freeze authority, Token-2022 extensions, holder concentration, dev wallet history, linked insider wallets, LP lock, honeypot sell test, liquidity, drawdown) and a dump-model probability for tokens under 6 h old. Accepts a mint, a pump fun / DexScreener link, or a $TICKER. |
| `scan_wallet` | Tokens in a public wallet, riskiest first. |
| `scan_link` | Wallet-drainer and phishing patterns in a URL or post. |
| `get_pro_quote` | Price, pay address, unique reference and order token for Pro. Moves no funds. |
| `verify_payment_and_get_key` | Verifies the payment on-chain and returns a Pro key. One signature, one key. |
| `get_pro_status` | Key validity and expiry, current limits. |
| `watch_token` / `unwatch_token` / `get_watch_status` | Pro agent watchlist (25 tokens): poll for dev or insider selling, liquidity pulls, holder jumps, failed sell tests, risk spikes. |

## Limits and Pro

Free: 5 checks a day per IP (shared between REST and MCP), the 3 riskiest tokens per wallet scan, 20 link scans.

Pro is $9 per 30 days in USDC or SOL on Solana, no auto-renewal: unlimited checks, full wallet scans, the agent watchlist, and Telegram holding alerts. An agent can buy it itself, with its user's approval:

1. `get_pro_quote` (or `GET /api/pro/quote`, or the body of the HTTP 402 returned past the free limit).
2. Send the payment in one transaction that carries the quote's `reference` (Solana Pay reference key or an SPL Memo containing it).
3. `verify_payment_and_get_key` with the signature and order token (or `POST /api/pro/verify`).

The REST API also implements x402 v2 (scheme `exact`, Solana mainnet USDC). It is advertised only while `https://rugradar-bgcm.onrender.com/api/pro/x402/status` reports `ready`.

## Links

- Agent docs: https://rugradar-bgcm.onrender.com/agents
- OpenAPI 3.1: https://rugradar-bgcm.onrender.com/openapi.json
- llms.txt: https://rugradar-bgcm.onrender.com/llms.txt
- Telegram bot: https://t.me/TopRugRadarBot
- X: https://x.com/TopRugRadar

## License

MIT. This repository contains only the stdio bridge and metadata; the RugRadar service itself is hosted.

TDQS

A4/5.0

Scored across 9 tools

Disambiguation5/5

Each tool targets a distinct resource and action: token checks, wallet scans, link scans, Pro status/quote/payment verification, and watchlist add/remove/status. Overlap is minimal, and the boundaries between check_token, scan_wallet, and scan_link are clear from their inputs and outputs.

Naming Consistency5/5

All names are snake_case and verb-led (get_, check_, scan_, watch_, unwatch_, verify_). The get_pro_* and watch_* families are especially predictable, and even the compound verify_payment_and_get_key follows a readable verb_noun style.

Tool Count5/5

Nine tools is well within a healthy range for this domain and each tool has a clear role: three scanners, three Pro subscription lifecycle tools, and three watchlist tools. Nothing feels redundant or missing at the count level.

Completeness4/5

The core surface is strong: token, wallet, and link risk scanning plus the full Pro purchase and watchlist lifecycle are covered. Minor gaps include no explicit standalone list/watchlist-listing tool and no batch multi-token check, but agents can work around these via get_watch_status or repeated calls.

Maintenance

ActivityMaintained
ResponsivenessNo recent community issues