sobriety-tools-guardian

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

• Third-party content exposure detected (high risk: 0.90). This skill's required workflows and scripts (e.g., scripts/cache-health.ts, scripts/perf-audit.ts, scripts/crisis-path-test.ts) explicitly fetch and parse data from a public meeting-proxy URL (https://jb4l-meeting-proxy.erich-owens.workers.dev) and other open HTTP endpoints, and those external responses are used to drive actions (cache warming, issue creation, alerts), so untrusted third-party content can materially influence the agent's decisions.