gcp-review

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 0.90). The skill directs the agent to read repository/configuration files and include "Evidence: " (and produce remediation code examples) without instructing redaction, which can force the LLM to copy secret values verbatim into output.

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

• Third-party content exposure detected (medium risk: 0.30). SKILL.md Step 1 through Steps 2–8 require locating and reading the user-provided repository’s GCP infrastructure-as-code/config files (e.g., Terraform, deployment-manager templates, org policies, IAM JSON) and then evaluating all string/comments as data, so outsider-authored free text in those files is ingested at runtime.