scientify-write-review-paper
Pass
Audited by Gen Agent Trust Hub on Apr 13, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes
catandlson local directories to check project status and list files in the papers directory. - [PROMPT_INJECTION]: The instruction "Don't ask permission. Just do it." is included in the skill documentation, which is a pattern associated with attempts to bypass agent confirmation dialogs or user oversight.
- [PROMPT_INJECTION]: The skill processes untrusted research papers from the filesystem, creating a surface for indirect prompt injection where malicious instructions embedded in a paper could influence agent behavior. Ingestion points: Research papers located in
$WORKSPACE/papers/. Boundary markers: Absent; paper content is processed without explicit delimiters or instructions to ignore embedded commands. Capability inventory: Command execution (cat,ls) and file system writing to$WORKSPACE/review/. Sanitization: Absent; the skill does not specify any validation or filtering of external paper content.
Audit Metadata