Skills
Navigate
skills/agentscope-ai/openjudge/02-bib-verify/Gen Agent Trust Hub

02-bib-verify

Pass

Audited by Gen Agent Trust Hub on Oct 25, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from external .bib files which could contain prompt injection attacks designed to influence the agent's interpretation of the verification results.
  • Ingestion points: The skill reads a .bib file specified by the user via the --bib_only flag (SKILL.md).
  • Boundary markers: No explicit boundary markers or instructions to ignore embedded instructions are provided in the skill guidelines.
  • Capability inventory: The skill performs file reading, network lookups to academic databases, and writes a markdown report to the local filesystem (SKILL.md).
  • Sanitization: There is no evidence of sanitization or filtering applied to the BibTeX content before processing.
  • [COMMAND_EXECUTION]: The skill instructs the agent to execute a Python module using python -m cookbooks.paper_review to process the data.
  • Evidence: The 'Quick start' and 'Relevant options' sections specify running shell commands to perform the verification task.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 25, 2026, 04:46 AM