NEAR Phishing Detector

• Risk level: Low
• Analyzed: Feb 20, 2026, 8:09 PM
• Findings: 0
• Version: 0.1.0

Overview

The skill's code, instructions, and resource requests are coherent with a NEAR phishing detection helper: it performs local heuristics against an embedded scam database and does not request credentials or reach out to external endpoints.

This skill appears to be a local heuristic detector with an embedded scam list and no external network calls or credential requirements. Before installing, consider: (1) Build and run the tests locally (npm install && npm run build && npm test) to confirm behavior in your environment; (2) Verify that your OpenClaw runtime will compile/serve the dist/ entrypoint as expected (SKILL.md references dist/index.js but the repo provides src/ with a build step); (3) Review and update the KNOWN_SCAMS list to match your threat model — the embedded database is the primary signal source; (4) If you plan to trust automated reports, cross-check results with official NEAR channels and moderators because heuristics can produce false positives/negatives; (5) If you will install npm dependencies on a production host, treat dependencies and package-lock as normal supply-chain risk and run dependency checks/lockfile verification.