What problem does it solve? Static scanners flag potential path traversal (CWE-22) and file upload (CWE-434) weaknesses in Spring Boot and JSP applications, but cannot prove they are actually exploitable. This Skill fires real payloads at a running staging or local target to confirm exploitability, eliminating false positives before remediation work begins. ## Core Features & Use Cases - Path Traversal Verification: Injects ../ variants (plain, URL-encoded, double-encoded, backslash, dotslash, null byte) into file parameters and confirms exploitation by detecting known file content signatures (/etc/passwd, web.xml, win.ini) in responses — read-only, non-destructive. - Upload Acceptance Testing: With an explicit --allow-destructive flag, uploads a harmless code-free marker file (.jsp extension containing only a nonce text) to verify whether dangerous extensions are accepted and retrievable from the webroot; never attempts code execution. - Fail-Closed Safety Gate: Every launch passes tools/scope_guard.py, which blocks production and public targets at the code level; credentials are masked in output and undetermined findings route to human confirmation via evidence_expectation cards. - Use Case: After a static scan flags a suspicious filePath parameter in a download endpoint, run this Skill against the local staging server to confirm whether ../../../etc/passwd actually leaks system file contents, then receive a severity-ranked report with the four-element remediation format. ## Quick Start Ask the AI to dynamically confirm path traversal and file upload vulnerabilities on your local staging app, for example: run the path traversal and upload exploit check against http://localhost:8080 using the flagged endpoints from my last scan.
