nature-data
Pass
Audited by Gen Agent Trust Hub on Aug 23, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user data, including manuscript notes, figure descriptions, and draft statements, which could potentially contain malicious instructions designed to influence the agent's output.
- Ingestion points: User-provided descriptions of data sources, figure/conclusion mappings, and initial availability drafts as specified in
SKILL.mdandREADME.md. - Boundary markers: The instructions do not define specific structural delimiters for user input, though they do specify logical boundaries for the agent's behavior, such as prohibiting the invention of DOIs or accession numbers.
- Capability inventory: The skill is limited to text analysis and generation of scientific documentation; it does not request tools for network access, file system modification, or code execution.
- Sanitization: No explicit sanitization or filtering of user-provided content is defined in the instructions.
Audit Metadata