The Agent Skills Directory

[DYNAMIC_EXECUTION]: The script scripts/lib/browser.mjs dynamically resolves and imports the playwright or playwright-core library from the project directory or global paths using createRequire().resolve() and import(). This is used to enable headless browser functionality if the dependency is present in the user's environment.\n- [INDIRECT_PROMPT_INJECTION]: The skill ingests and processes untrusted HTML and metadata from external URLs through various audit scripts. \n
Ingestion points: External site content is fetched using fetch or Playwright in scripts/lib/pages.mjs.\n
Boundary markers: Findings are labeled with evidence types (e.g., OBSERVED, SOURCE-INDICATED) in the final report to distinguish site data from analysis results.\n
Capability inventory: The skill performs network requests, reads local files for secret scanning, and executes git commands.\n
Sanitization: While the audit scripts return structured data, raw content from the target site (such as titles, text, or form attributes) is included in the diagnostic output.\n- [COMMAND_EXECUTION]: The script scripts/scan-secrets.mjs executes the git command using execFileSync to determine which files are tracked in the repository.

production-all