conflict-detection

Pass

Audited by Gen Agent Trust Hub on Feb 16, 2026

Risk Level: LOWSAFE
Full Analysis
  • [Prompt Injection] (SAFE): No patterns detected that attempt to override system instructions or bypass safety filters.
  • [Data Exposure & Exfiltration] (SAFE): No hardcoded credentials or sensitive data access. File paths like ITEM-XXX.md are used for local project state.
  • [Remote Code Execution] (SAFE): No mechanisms for downloading or running external scripts or packages exist.
  • [Indirect Prompt Injection] (LOW): The skill processes external data to influence reasoning and local state. 1. Ingestion: User input and ITEM-XXX.md. 2. Boundary markers: Absent. 3. Capability inventory: Local file modification, no shell or network access. 4. Sanitization: Absent.
  • [Obfuscation] (SAFE): No hidden or encoded content was found during the analysis.
Audit Metadata
Risk Level
LOW
Analyzed
Feb 16, 2026, 06:51 AM