temper

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

• Third-party content exposure detected (high risk: 1.00). The skill explicitly fetches PR metadata (title/body) from third-party forges via the gh/glab/bb CLI (see "Step 1: Resolve the review scope" in SKILL.md) and injects that untrusted, user-generated PR body into the reviewer template as {PLAN_REFERENCE}, which the agent is expected to read and use to drive review decisions and next actions.