The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill facilitates the installation of the Chromium browser engine required for report rendering.
Detailed evidence: The setup.sh script executes npx playwright install chromium to fetch the browser binaries from the official Playwright (Microsoft) registry.
[INDIRECT_PROMPT_INJECTION]: The skill processes data from external files, creating a surface for indirect prompt injection through ingested audit findings.
Ingestion points: External data is ingested from findings.json (audit output) and report.json (user configuration) within scripts/prepare.py.
Boundary markers: Absent; the skill does not wrap processed content in specific delimiters before rendering into the PDF template.
Capability inventory: The skill enables local shell command execution (via preparation and build scripts) and persistent file system writes (to output the PDF).
Sanitization: Partial sanitization is implemented in scripts/prepare.py, which replaces sensitive terms like 'attacker' with neutral equivalents. While the template engine in build.mjs escapes standard values, it permits raw HTML injection via the {{{html}}} syntax in templates/document.html for formatting specific report sections, assuming the input source is trusted.

create-security-report