paper-review

Pass

Audited by Gen Agent Trust Hub on May 12, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill uses uvx and uv run to download and execute opencite, pdf2image, and pillow at runtime. These are used for manuscript conversion and literature search.
  • [COMMAND_EXECUTION]: Shell commands and Python one-liners are executed for PDF processing. These are localized to the task of converting documents for the agent's review.
  • [PROMPT_INJECTION]: The skill processes untrusted PDF manuscripts, creating an indirect prompt injection surface.
  • Ingestion points: manuscript.pdf (in SKILL.md) is converted for analysis.
  • Boundary markers: None provided in the ingestion logic.
  • Capability inventory: Shell and Python execution for file manipulation.
  • Sanitization: No validation of PDF content is performed.
Audit Metadata
Risk Level
SAFE
Analyzed
May 12, 2026, 01:35 PM
Security Audit — agent-trust-hub — paper-review