venue-guidelines-compliance
Pass
Audited by Gen Agent Trust Hub on Aug 24, 2026
Risk Level: SAFEPROMPT_INJECTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
- [PROMPT_INJECTION]: The skill processes untrusted data from external URLs and paper content, creating an indirect prompt injection surface.
- Ingestion points: External URLs via --guidelines and paper TeX/PDF content.
- Boundary markers: Step 2 includes instructions to ignore embedded instructions.
- Capability inventory: Execution of latex, qpdf, and texcount; local file system write access to reviews/.
- Sanitization: Relies on Natural Language guidance for the agent rather than structural filtering or sandboxing.
- [EXTERNAL_DOWNLOADS]: The skill fetches guidelines from external sources provided via CLI arguments to compare against the active paper.
- [COMMAND_EXECUTION]: The skill executes several command-line tools for PDF verification (qpdf), word counting (texcount), and document compilation (latex), as well as a bash script for logging review results.
Audit Metadata