The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill provides instructions to clone a repository from https://github.com/ccyl13/Pentestify.git, which is an external source.
[REMOTE_CODE_EXECUTION]: The manual installation steps guide the user to execute the cloned code using python3 run.py, leading to the execution of unverified external scripts.
[COMMAND_EXECUTION]: The installation process involves executing multiple shell commands, including git clone, pip install, and playwright install chromium, which can modify the system environment.
[PROMPT_INJECTION]: The skill exhibits a vulnerability surface for indirect prompt injection within the PDF generation logic described in SKILL.md.
Ingestion points: User-provided data in report findings (titles, descriptions, impact, and evidence) is processed during PDF export.
Boundary markers: There are no delimiters or instructions provided to the agent to ignore embedded commands or scripts within the findings data.
Capability inventory: The skill utilizes playwright for rendering, FastAPI for network operations, and SQLite for local file persistence.
Sanitization: The generate_pdf_custom code example performs direct string concatenation of report_html into the Playwright content setter without escaping or validation, creating an XSS-to-PDF or SSRF injection surface.

pentestify-security-report-generator