reducing-aigc-detection

Pass

Audited by Gen Agent Trust Hub on Jun 16, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill facilitates Indirect Prompt Injection by processing untrusted external documents (DOCX, PDF) using high-privilege tools.
  • Ingestion points: External files are read into the agent's context during the Phase 0 reconnaissance.
  • Boundary markers: Instructions do not define the use of delimiters to isolate untrusted document content from agent instructions.
  • Capability inventory: The agent has access to Bash, Write, and Edit, creating a risk if malicious instructions are hidden in target papers.
  • Sanitization: No validation or sanitization of input text is required by the skill logic.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 16, 2026, 06:20 PM
Security Audit — agent-trust-hub — reducing-aigc-detection