team-code-review

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

• Third-party content exposure detected (high risk: 0.75). The workflow instructs the agent to “read the repository instructions first” and “inspect the complete diff and relevant surrounding code,” which—when the target is “another contributor’s pull request”—means the LLM will ingest outsider-authored PR text/diff content into its context for review.