The Agent Skills Directory

[COMMAND_EXECUTION]: The skill includes a utility scripts/lib/browser.mjs that identifies and executes local Chromium or Chrome binaries using Playwright to render web pages for analysis.
[COMMAND_EXECUTION]: The scripts/scan-secrets.mjs script utilizes execFileSync to run the git command locally, which is used to identify tracked files and verify repository status.
[INDIRECT_PROMPT_INJECTION]: The skill possesses an attack surface for indirect prompt injection because it fetches and processes external content (HTML, sitemaps, robots.txt) from the sites it audits.
Ingestion points: Content is fetched via fetch or read from the local filesystem in scripts/lib/pages.mjs.
Boundary markers: Results are formatted into structured reports using scripts/lib/report.mjs before being presented to the agent.
Capability inventory: The skill can perform network requests (fetch), execute shell commands (git), and drive a headless browser (playwright).
Sanitization: Content is parsed using a custom, non-executing HTML parser, though rendering with Playwright involves executing client-side JavaScript from the target site.

deployment-cleanup