perses

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

• Third-party content exposure detected (high risk: 1.00). The skill's required workflows explicitly fetch and ingest third-party dashboards and API responses (e.g., references/migration.md Phase 1 EXPORT uses curl to "https://grafana.example.com/api/..." and references/dashboard.md Phase 1 FETCH uses perses_get_dashboard_by_name or percli describe dashboard) and then parse/analyze and potentially fix or deploy based on that content, so untrusted external/user-generated content can directly influence agent actions.