ScanHood: open-source safety layer

This is the actual code running live behind scanhood.xyz, a safety and analytics layer for Robinhood Chain (chain ID 4663). Published so the detection logic can be read and audited directly, instead of trusted as a black box.

No external security API covers this chain (GoPlus rejects chain 4663, GMGN's API is walled off), so every check here runs on-chain, from scratch.

What's in here

• honeypot/scan.js: the honeypot/rug detector. Simulates a real buy + sell round-trip via a single eth_call (no gas spent, nothing deployed): it constructs a small contract whose constructor buys the token, tries to sell it, and reverts with the encoded result. That revert-with-data trick is the whole mechanism, read the file, it's under 150 lines.
• api/server.js: the REST API (/api/scan, /api/search, /api/ohlcv, /api/quote, /api/launch, /api/watchlist, /api/token-extra, the /api/udf/* TradingView datafeed, and the opt-in fee-claim bot's /api/bots/*). Composes the honeypot result with an LP-lock check (who actually controls the liquidity), contract-verification status, and deployer-reputation lookup into one PASS / CAUTION / DANGER verdict.
• mcp/server.js: a Model Context Protocol server exposing the same API as native tools, so an agent (Claude, etc.) can call scan_token, safe_launches, build_launch_tx, etc. directly instead of hand-rolling HTTP calls.

Live docs with request/response examples: https://scanhood.xyz/docs

Architecture, honestly

            ┌─────────────┐
  agents ──▶│ mcp/server  │──▶  api/server  ──▶  honeypot/scan (eth_call sim)
  humans ──▶│  (MCP)      │        │        ├──▶ Blockscout (verification, LP holders)
            └─────────────┘        │        └──▶ DexScreener (price/liquidity/volume)
                                    ▼
                    optional data feeds (deployer reputation,
                    other-launchpad allowlists, RWA issuer set,
                    each is a separate cron-refreshed collector,
                    not included in this repo, degrades gracefully
                    to "unknown" if absent)

Everything marked "optional" in .env.example is a static JSON file your own cron/collector can produce in whatever shape matches the field names read in api/server.js, the API itself never assumes a specific pipeline for them.

Run it

npm install
cp .env.example .env   # optional, every feature has a working default
npm run api            # http://127.0.0.1:8946
npm run mcp             # http://127.0.0.1:8948/mcp
node honeypot/scan.js 0xYourTokenAddress   # CLI, standalone

What's not in here, and why

• The launchpad contracts. ScanHood's no-rug launchpad is a fork of the open-source Pons locked-LP factory, that's upstream's code to publish, not ours to re-vendor here. api/server.js's /api/launch just needs a factory address + fee in .env; point it at any factory with the same launchToken/predictTokenAddress signature.
• The frontend. This repo is the audit surface (the "how do we decide PASS/CAUTION/DANGER" logic), not the site's UI.
• Trade execution / custody. Nothing here holds a private key. /api/launch and /api/quote only ever return unsigned transactions or read-only quotes for you to sign yourself.

Disclaimer

These are automated on-chain heuristics. They reduce risk; they are not a guarantee. "Sellable" means the honeypot simulation passed, not "safe to buy." Always do your own research.