file-reference

Pass

Audited by Gen Agent Trust Hub on Feb 16, 2026

Risk Level: LOWNO_CODEPROMPT_INJECTION
Full Analysis
  • Indirect Prompt Injection (LOW): The skill is designed to ingest and interpret external file content, which is a common vector for indirect prompt injection attacks where a file contains instructions to divert the AI's behavior.
  • Ingestion points: Uses the Read tool to access files based on user references (@name, natural language).
  • Boundary markers: Absent. The prompt does not establish clear delimiters or instructions for the agent to ignore commands found within the analyzed files.
  • Capability inventory: The skill is restricted to the Read tool and does not possess file-writing, network access, or shell execution capabilities.
  • Sanitization: No sanitization or content validation steps are specified for the ingested file data.
Audit Metadata
Risk Level
LOW
Analyzed
Feb 16, 2026, 09:43 AM