presentation-builder

Pass

Audited by Gen Agent Trust Hub on Nov 9, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill relies exclusively on standard Node.js built-in modules (fs, path, child_process, vm, os) for its operations, minimizing supply chain risks.
  • [SAFE]: The scripts/create-presentation.cjs utility implements comprehensive sanitization for both user-supplied text and author-supplied HTML. Functions like escapeHtml and sanitizeAuthorHtml proactively remove dangerous elements such as <script>, <style>, <iframe>, and event handlers (e.g., onclick), as well as javascript: URI schemes.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted source material to generate slides, which constitutes an indirect injection surface.
  • Ingestion points: Untrusted data enters via the "source material" processed by the agent in SKILL.md and the JSON specification ingested by scripts/create-presentation.cjs.
  • Boundary markers: The instructions require the agent to maintain a "source ledger" and a "design-authority record" to ground every claim in verifiable evidence.
  • Capability inventory: The skill can write files to the local file system using fs.writeFileSync in scripts/create-presentation.cjs.
  • Sanitization: Multiple sanitization layers are implemented to strip executable code and validate URL safety within the generated HTML.
  • [DYNAMIC_EXECUTION]: The test suite (tests/create-presentation.test.cjs) uses the node:vm module to execute and verify the logic of the generated JavaScript runtime within a isolated context. This is a standard and controlled use of dynamic execution for unit testing.
  • [OBFUSCATION]: The file tests/skill-genericity.test.cjs uses String.fromCharCode to generate a string for checking against subject-specific coupling. This is a benign use of encoding for automated testing of project-neutrality and does not conceal malicious code.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 9, 2026, 01:34 AM