check-upstream-flake

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

• Third-party content exposure detected (high risk: 0.70). The skill directly queries the public LUCI Analysis REST API at analysis.api.luci.app to fetch test history and flakiness data and then interprets those results to produce verdicts and filter recommendations, meaning untrusted third-party content can materially influence the agent's decisions.