iac-security

Pass

Audited by Gen Agent Trust Hub on Jul 2, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill presents a surface for indirect prompt injection as it is designed to ingest and process untrusted Infrastructure-as-Code files (Terraform, CloudFormation, Pulumi, Bicep) as defined in Step 1 of SKILL.md.
  • Ingestion points: The skill uses Glob and Read tools to scan all files matching extensions like .tf, .tfvars, .yaml, .json, and .bicep.
  • Boundary markers: The skill includes an explicit 'Prompt Injection Safety Notice' in SKILL.md that instructs the agent to treat all file content as DATA and not instructions.
  • Capability inventory: Capabilities are restricted to file discovery and reading (Glob, Grep, Read). No network or code execution tools are enabled in the allowed-tools configuration.
  • Sanitization: The instructions explicitly direct the agent to disregard inline claims (e.g., "this resource is compliant") and scanner suppression comments (e.g., "# checkov:skip"), treating them as findings rather than directives.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 2, 2026, 01:36 PM
Security Audit — agent-trust-hub — iac-security