The Agent Skills Directory

[SAFE]: The skill operates as a set of diagnostic tools for web development. All script activities, including network requests for link validation and local file system access for scanning project directories, are consistent with its stated purpose of identifying launch readiness issues.
[DYNAMIC_EXECUTION]: The skill uses dynamic imports in scripts/lib/browser.mjs to load the playwright library. This is a standard pattern for optional dependencies, allowing the tool to run in static mode if the browser automation library is not present. The paths are resolved through the standard Node.js resolution mechanism and are restricted to specific, well-known libraries.
[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from external websites for analysis.
Ingestion points: scripts/lib/pages.mjs fetches HTML content from user-supplied URLs via fetchWithRetry or renderPages.
Boundary markers: The skill uses a structured reporting format defined in scripts/lib/report.mjs to separate tool output from instructions, though explicit "ignore embedded instructions" warnings are absent.
Capability inventory: The skill can perform network operations via scripts/check-links.mjs (fetch) and execute browser instances via scripts/lib/browser.mjs.
Sanitization: The skill uses a custom HTML parser in scripts/lib/html.mjs to extract text and attributes but does not perform sanitization of external content intended for LLM safety.
[EXTERNAL_DOWNLOADS]: The scripts/check-links.mjs tool performs network requests to verify the status of external links found on the audited website. These requests are limited to HTTP HEAD/GET calls for status code verification and do not involve downloading or executing remote code payloads.

trust-all