academic-ref-check

Pass

Audited by Gen Agent Trust Hub on Aug 27, 2026

Risk Level: SAFE
Full Analysis
  • [DATA_EXPOSURE_AND_EXFILTRATION]: The skill transmits bibliographic metadata (titles, DOIs, authors) to api.openalex.org and api.crossref.org for verification. These are established, well-known academic metadata providers. No sensitive system files, environment variables, or user credentials are accessed or transmitted.
  • [EXTERNAL_DOWNLOADS]: The skill documentation specifies dependencies on standard, reputable Python libraries including python-docx for Word document generation and requests for network communication.
  • [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process user-provided bibliography files (Markdown and JSON). While this presents a data ingestion surface, the provided scripts use robust parsing via regular expressions and standard JSON loaders, minimizing the risk of the agent executing instructions embedded within the processed citations.
  • [COMMAND_EXECUTION]: The instructions involve executing local Python scripts provided within the skill for document conversion and verification. These scripts do not take arbitrary shell input or perform dangerous subprocess operations.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 27, 2026, 11:00 AM
Security Audit — agent-trust-hub — academic-ref-check