reference-manager

Pass

Audited by Gen Agent Trust Hub on Jun 16, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill operates strictly on local files within the project workspace (e.g., paper/sections/, workspace/papers/) to extract metadata and verify citations.
  • [SAFE]: There are no network operations, remote code executions, or external dependencies detected.
  • [SAFE]: The instructions include robust anti-hallucination rules, specifically forbidding the fabrication of DOIs, author names, or titles, and requiring explicit flagging of unverified sources.
  • [SAFE]: The skill does not request or use privileged commands (e.g., sudo, chmod) or persistent mechanisms.
  • [SAFE]: While the skill processes untrusted user data (draft sections and PDFs), its capabilities are limited to writing structured BibTeX files and markdown audit reports, posing a low risk of indirect prompt injection.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 16, 2026, 05:24 AM
Security Audit — agent-trust-hub — reference-manager