antislop-decks
Pass
Audited by Gen Agent Trust Hub on Sep 13, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest external configuration files like
DESIGN.mdand existing presentation templates to ensure brand consistency. This behavior, while intended, creates a standard surface for indirect prompt injection if the source files are untrusted. - Ingestion points: Processes
DESIGN.mdfiles from the project directory and reads existing.pptxstructures via OOXML unpacking. - Boundary markers: No specific delimiters or "ignore previous instructions" guards are established for the ingested design tokens or content.
- Capability inventory: The skill instructions involve writing to the file system (generating
DECK.mdand presentation files) and interacting with presentation APIs (Google Slides). - Sanitization: The skill does not provide explicit instructions for sanitizing the narrative content or design specifications before they are interpolated into the final output.
Audit Metadata