ieee-parse-results

Pass

Audited by Gen Agent Trust Hub on Mar 15, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill parses content from an external website (IEEE Xplore). This introduces a surface for indirect prompt injection, where malicious text on the webpage could attempt to influence the agent's instructions.
  • Ingestion points: SKILL.md (extracts DOM content via evaluate_script)
  • Boundary markers: Absent
  • Capability inventory: None (the script only reads and returns data)
  • Sanitization: Basic regex used to strip HTML tags from paper titles.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 15, 2026, 07:11 AM