rhdh-pr-review

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

• Third-party content exposure detected (medium risk: 0.30). In scripts/fetch_pr_context.py used by workflows/fetch-github.md, the agent fetches GitHub PR bodies/diffs and also fetches linkedIssues (issue titles/bodies) and existingComments/existingReviews via gh api for the specified PR, thereby ingesting outsider-authored free text from those PR/issue/comment sources at runtime.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.90). The workflows fetch and apply remote install manifests and CI-built container images at runtime (e.g., curl "https://raw.githubusercontent.com/redhat-developer/rhdh-operator/${PR_BRANCH}/dist/rhdh/install.yaml" and docker://quay.io/rhdh-community/operator:TAG), which causes execution of external code on the target cluster and the skill depends on those resources.