{
  "claim_outcome": "confirmed",
  "claim_block_reason": null,
  "repro_result": "confirmed",
  "validated_surface": "api_remote",
  "evidence_scope": "production_path",
  "claimed_impact_class": "code_execution",
  "observed_impact_class": "code_execution",
  "exploitability_confidence": "high",
  "attacker_controlled_input": "unauthenticated workflow definition with an INLINE task whose JavaScript expression bootstraps Java reflection from the bound input object ($) to java.lang.Runtime and calls Runtime.exec(['sh','-c',CMD])",
  "trigger_path": "POST /api/metadata/workflow (register malicious workflow, no auth) + POST /api/workflow/{name} (start, no auth) -> INLINE system task evaluates the JS expression in a GraalVM Context built with HostAccess.ALL -> reflective Runtime.exec -> command stdout returned in GET /api/workflow/{id}?includeTasks=true tasks[].outputData.result",
  "end_to_end_target_reached": true,
  "sanitizer_used": false,
  "crash_observed": false,
  "read_write_primitive_observed": false,
  "exploit_chain_demonstrated": true,
  "blocking_mitigation": null,
  "inferred": false
}