papercraft-stop-motion-explainer
Warn
Audited by Gen Agent Trust Hub on Nov 2, 2026
Risk Level: MEDIUMMETADATA_POISONINGINDIRECT_PROMPT_INJECTION
Full Analysis
- [METADATA_POISONING]: The skill metadata in 'meta.yaml' identifies the author as 'MiniMax Hub' and marks the source as 'official'. This contradicts the actual developer attribution, which could mislead users into trusting the skill as a verified first-party component.\n- [INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied 'source material' and 'knowledge points' to generate creative directions, prompts, and storyboards. The instructions lack specific boundary markers or sanitization requirements, creating a surface for potential injection attacks where instructions hidden in the input material could influence agent behavior.\n
- Ingestion points: SKILL.md (Step 1: Understand the Input)\n
- Boundary markers: Absent; the agent is instructed to use user topic and domain words directly without delimiters.\n
- Capability inventory: Writing multi-section planning documents to canvas, generating prompts for MiniMax-H3 video model, and audio generation/repair (Step 17).\n
- Sanitization: Absent; no validation or escaping of the user's input material is specified.
Audit Metadata