pentestify-security-report-generator
Fail
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: HIGHEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTIONDATA_EXFILTRATION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill instructs the agent to clone source code from an unverified GitHub repository (https://github.com/ccyl13/Pentestify.git).
- [REMOTE_CODE_EXECUTION]: The skill executes code from the unverified external repository using python3 run.py and installs potentially untrusted dependencies via pip.
- [DYNAMIC_EXECUTION]: The backend logic for PDF generation uses playwright to render raw HTML strings constructed by concatenating user-supplied report data with CSS, which can lead to script injection or server-side request forgery (SSRF).
- [INDIRECT_PROMPT_INJECTION]: The skill creates a vulnerability surface by (1) ingesting untrusted security findings via POST /api/reports/{report_id}/findings in SKILL.md, (2) failing to use boundary markers or delimiters for these inputs, (3) possessing capabilities such as PDF generation via Playwright and database access, and (4) lacking explicit sanitization or validation of the ingested payloads before rendering.
- [DATA_EXFILTRATION]: The skill implements endpoints (/api/database/export) that allow for the direct extraction of the underlying SQLite database containing all security findings.
Recommendations
- AI detected serious security threats
Audit Metadata