iac-security

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The skill explicitly requires including "Evidence: " in findings and instructs scanning state files (e.g., terraform.tfstate), which would cause the LLM to reproduce literal code lines that may contain secrets verbatim.

CRITICAL E005: Suspicious download URL detected in skill instructions.

• Suspicious download URL detected (high risk: 0.80). All listed links are reputable documentation or examples except the git module URL pinned to a branch (ref=main), which is mutable/unpinned and represents a supply-chain risk that could be used to distribute malicious code.

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (medium risk: 0.65). SKILL.md describes scanning IaC files (terraform/cloudformation/pulumi/bicep) and compiling findings that include evidence/strings from those templates; at runtime, those templates are commonly outsider-authored (e.g., PR/issue/wiki content or provided IaC from an external repo) and their free-form text (comments/descriptions/tag values/inlined strings) can be ingested into the LLM context as part of the “Detailed Findings”/evidence.