config-scan

Pass

Audited by Gen Agent Trust Hub on Sep 15, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill's primary function is to ingest and analyze untrusted configuration files which could theoretically contain malicious instructions.
  • Ingestion points: Processes user-provided .env files, Dockerfiles, Kubernetes manifests, Terraform code, and application configuration files (JSON, YAML, JS).
  • Boundary markers: No specific delimiters or "ignore embedded instructions" markers are defined for the data being analyzed.
  • Capability inventory: The skill contains no executable scripts, subprocess calls, or network operations; it provides analysis patterns for the agent to use.
  • Sanitization: No explicit sanitization or validation of the external configuration data is mentioned.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 15, 2026, 05:47 PM
Security Audit — agent-trust-hub — config-scan