The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses spawnSync in scripts/run.mjs to execute its internal TypeScript components using the tsx runtime resolved from the target project. This is a standard execution pattern for project-local tooling.
[EXTERNAL_DOWNLOADS]: The snapshot.ts and extract_inline_html.ts scripts fetch remote images and stylesheets to inline them into the static output. The implementation includes a robust SSRF protection layer in scripts/extract_inline_html.ts that validates DNS answers and blocks requests to non-public (local, reserved, or private) IP addresses.
[DYNAMIC_EXECUTION]: The skill employs dynamic loading techniques in several places:
scripts/project_dependencies.mjs resolves and requires dependencies like puppeteer and @babel/parser from the target project at runtime.
scripts/snapshot.ts allows users to provide an optional authentication script via the --auth-script flag, which is loaded using dynamic import() to perform interactions before page capture.
[INDIRECT_PROMPT_INJECTION]: The skill processes potentially untrusted data from local JSX files and external URLs. While this presents an indirect prompt injection surface, the risk is mitigated by:
Ingestion points: snapshot.ts (external URLs) and extract_inline_html.ts (local source files).
Boundary markers: Explicit instructions in SKILL.md and the Rules section that emphasize evidence-based reporting and prohibit data fabrication.
Capability inventory: The skill possesses file writing, network access (specifically for asset fetching), and subprocess execution capabilities.
Sanitization: Implementation of escapeHtml and escapeAttribute in scripts/extract_inline_html.ts and log sanitization via scripts/safe_diagnostics.ts to prevent credential leakage and injection.

stitch-extract-static-html