What problem does it solve? Security reviews of Node.js backends often degrade into generic OWASP checklists or vague hardening advice that misses the actual exploitable issue. This Skill produces findings-first reviews that name the exact broken trust boundary, trace the shortest attacker path, and recommend the smallest safe fix. ## Core Features & Use Cases - Findings-First Review: Identifies concrete security findings in auth, session, cookie, secret-handling, outbound HTTP, and SSRF surfaces with severity calibration from Blocker to Low. - Fail-Open Detection: Catches verification failures, missing secrets, and error paths that silently downgrade to allow or guest access instead of denying. - Multiple Review Modes: Supports diff/PR review, audit mode for unfamiliar backends, and incident/exploit reconstruction. - Use Case: A developer submits a Fastify PR that decodes a JWT before verifying it and falls back to a guest role on verification errors. The Skill flags both as fail-open auth findings, traces the forged-token exploit path, and recommends verifying before trusting claims. ## Quick Start Ask the AI to do a findings-first security review of your Fastify auth middleware or outbound webhook code and list only the highest-signal findings.