ig-carousel-writer
Pass
Audited by Gen Agent Trust Hub on Sep 19, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONNO_CODE
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data provided by the user to generate social media content, creating a potential surface for embedded instructions in the source text.
- Ingestion points: Processes user-supplied primary drafts, transcripts, and source materials as described in SKILL.md and '如何使用.md'.
- Boundary markers: Does not utilize explicit data delimiters but enforces strict adherence to verified source facts and brand templates to limit model deviation.
- Capability inventory: Instructed to perform file system writes to specific project directories and utilize available image generation tools (e.g., imagegen).
- Sanitization: Explicitly forbids the generation of fictitious facts, stories, or external calls-to-action (CTAs) not present in the verified source material, acting as a functional filter against content-based injection.
Audit Metadata