security-docker

Pass

Audited by Gen Agent Trust Hub on Feb 17, 2026

Risk Level: SAFE
Full Analysis
  • [Indirect Prompt Injection] (LOW): The skill analyzes untrusted configuration files which creates a potential surface for indirect prompt injection.
  • Ingestion points: scripts/scan.sh reads contents of Dockerfile and docker-compose.yml files using ripgrep and grep.
  • Boundary markers: No explicit delimiters or instructions to ignore embedded content are present in the script or prompt logic.
  • Capability inventory: scripts/scan.sh performs local file reading and pattern matching but does not execute external code from the files.
  • Sanitization: No sanitization or escaping of ingested file content is performed during the audit process.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 17, 2026, 06:12 PM