@solsentry/guard

Pre-sign transaction guard for Solana wallets and dapps, backed by the SolSentry threat-intelligence API.

Ask one question before your user signs: is anything in this transaction a known threat? The SDK extracts the programs a transaction touches, checks them against SolSentry's operator-risk data, screens recipient addresses for address-poisoning lookalikes, and decodes common signature-phishing vectors — then returns a single worst-case verdict: safe · unknown · caution · dangerous.

Free public API underneath — no API key required. Every risk verdict is auditable per-mint at https://api.solsentry.app/v1/predictions/{mint}.

Install

npm install @solsentry/guard

Quickstart

import { SolSentryGuard } from "@solsentry/guard";

const guard = new SolSentryGuard();

// Before your wallet signs anything:
const advice = await guard.analyzeBeforeSign(transactionBytes);
if (advice.verdict === "dangerous") {
  // block, or show the findings to the user
  console.warn(advice.findings);
}

Individual checks

// Program safety (is this program associated with drains/rugs?)
const report = await guard.analyzeProgram("675kPX9MHTjS2zt1qfr1NYHuzeLXfQM9H24wFSUt1Mp8");

// Address-poisoning screen (lookalike of an address you've seen before?)
const look = await guard.checkLookalike(candidate, knownAddresses);

// Signature-phishing vector decode (no network call — pure local heuristics)
import { decodeSignaturePhishing } from "@solsentry/guard/phishing";
const warnings = decodeSignaturePhishing({ instructions });

What it is / isn't

• It's a thin client over the free SolSentry REST API plus local heuristics — auditable, no magic.
• It is not a guarantee. Verdicts reflect SolSentry's live operator-risk data; new operators start as unknown. Treat dangerous as a hard block and caution as "show the user why".

License

MIT