{
  "paper": "2606.26859",
  "method": "agentx",
  "benchmark": "planbench-mini",
  "seed": 42,
  "episodes": 120,
  "metrics": {
    "answer_accuracy": 1.0,
    "plan_success": 1.0,
    "joint_success": 1.0,
    "average_cost": 0.5869999999999987
  },
  "diagnostics": {
    "skills_created": 12,
    "skills_reused": 108,
    "plan_explorations": 36,
    "local_verifier_calls": 132,
    "global_verifier_calls": 120,
    "archival_writes": 12,
    "policy_updates": 120,
    "fidelity": "mechanism reproduction on deterministic benchmark mini-suite"
  },
  "diagnostic_only": true,
  "schema_version": 2,
  "manifest_ref": "agent-research:planbench-mini-seed42",
  "evaluation_protocol": {
    "tier": "l1_mechanism",
    "seeds": [
      42
    ],
    "formal_comparison": false,
    "claim_policy": "single/few-seed smoke result; do not claim a stable improvement"
  },
  "provenance": {
    "artifact_path": "docs/agent-research/2606.26859-agentx/metrics/planbench-mini-seed42.json",
    "historical_migration": "historical-metrics-v2-2026-08-09",
    "original_code_commit": "not recorded",
    "dataset_fingerprint": "not recorded in historical artifact"
  }
}