science-illustration-skill
Pass
Audited by Gen Agent Trust Hub on Oct 18, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied research materials (abstracts, methods, captions) and interpolates this content into image generation prompts.
- Ingestion points: Workflow Step 1 in
SKILL.mddescribes extracting entities and labels from user-provided source material. - Boundary markers: The prompt template in
references/prompt-template.mduses bracketed placeholders (e.g.,{topic},{message}) without explicit delimiters or instructions to ignore nested commands within the user data. - Capability inventory: The skill utilizes the
image_gentool to produce visuals. - Sanitization: No explicit sanitization or filtering of user input is mentioned, though the skill contains strict internal checks in
references/scientific-integrity.mdto prevent the generation of fabricated evidence.
Audit Metadata