slide-deck-builder
Pass
Audited by Gen Agent Trust Hub on Aug 31, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes local auditing scripts (
audit-html-rendered.mjsandaudit-pptx.py) using Node.js and Python to validate generated content. - [EXTERNAL_DOWNLOADS]: The skill uses the GitHub CLI to fetch the contribution guide from the author's official repository (
opengovsg/corp-ai-skills). - [DYNAMIC_EXECUTION]: The skill validates HTML slide decks by rendering them in a headless Chrome/Chromium browser instance through the
audit-html-rendered.mjsscript, which injects a validation runner into the file at runtime. - [INDIRECT_PROMPT_INJECTION]: The skill possesses a surface for indirect prompt injection as it renders agent-generated HTML in a browser and parses review files for quality auditing.
- Ingestion points: User-supplied presentation content and
deck-review.mdfiles. - Boundary markers: Absent; user-influenced content is rendered directly to perform visual checks.
- Capability inventory: Subprocess execution of headless Chrome, Python, and Node.js; file system read/write access.
- Sanitization: Absent; content is processed and rendered to validate layout and contrast.
Audit Metadata