What problem does it solve? Installing AI agent skills from untrusted sources can introduce malicious code, prompt injection, credential theft, or data exfiltration into your environment. This Skill performs static security audits on skill directories or git repos before installation, producing a clear PASS, WARN, or FAIL verdict. ## Core Features & Use Cases - Code Pattern Scanning: Detects dangerous patterns in Python, Bash, and JavaScript files such as os.system, eval, subprocess with shell=True, base64 obfuscation, network exfiltration, and credential harvesting. - Prompt Injection Detection: Scans SKILL.md and markdown references for system prompt overrides, role hijacking, safety bypass directives, hidden zero-width characters, and data extraction instructions. - Supply Chain & Filesystem Checks: Flags typosquatted dependencies, unpinned versions, runtime pip installs, hidden files, binaries, symlinks, and writes outside the skill boundary. - Use Case: Before installing a community skill from GitHub, run the auditor against the repo URL to receive a severity-grouped report with remediation guidance for each finding. ## Quick Start Audit a local skill directory by asking the agent to run the skill_security_auditor.py script against the target path and report the verdict.
