What problem does it solve? Web applications that print user input into HTML or JavaScript contexts without escaping are exposed to Cross-Site Scripting (XSS), and manual code review across mixed Spring Boot, legacy JSP, and Vue/Nuxt frontends easily misses these sinks. This Skill statically scans source code for XSS candidates and then applies AI context verification to remove false positives and confirm real vulnerabilities. ## Core Features & Use Cases - Stack-aware scanning: Auto-detects spring-modern, jsp-legacy, and frontend (Vue/Nuxt) stacks per directory and applies matching Semgrep rules, with a grep fallback when Semgrep is unavailable. - AI false-positive filtering: Verifies each candidate against escaping patterns such as fn:escapeXml, c:out, th:text, textContent, and DOMPurify sanitization so safe code is excluded. - Four-element reporting: Every confirmed finding documents what is vulnerable, why, how it is exploited (conceptual PoC), and how to fix it, ordered by severity. - Use Case: Point it at a legacy JSP bulletin board project to find stored XSS in ${board.bbsContent} output, then cross-check the Vue frontend's v-html bindings against backend sanitization logic. ## Quick Start Ask the AI to scan your project for XSS vulnerabilities by running the scan_xss.py script against your source directory and verifying the candidates.