task-routing

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

• Third-party content exposure detected (high risk: 0.80). The skill explicitly runs and parses GitHub data (e.g., "gh issue list -R {target_repo}" and "gh project item-list ... | python3 ..." in SKILL.md Step 3, and "gh label list -R {target_repo}" in Step 4), which ingests user-generated, potentially public issue and project content that the agent reads and uses to decide whether to update or create issues—so untrusted third-party content can materially influence actions.