{
  "schema_version": 2,
  "method": "skill-state",
  "manifest_ref": "agent-research:skill-state",
  "seeds": [
    42,
    43,
    44
  ],
  "dataset": {
    "id": "synthetic-skill-state-public-counter",
    "revision": "oct19-skill-state-contract-v1",
    "capability_benchmark": false
  },
  "evaluation_tier": "l1_mechanism",
  "evaluation_protocol": {
    "tier": "l1_mechanism",
    "seeds": [
      42,
      43,
      44
    ],
    "formal_comparison": false,
    "claim_policy": "bounded component/protocol fixtures; not capability or official benchmark evidence"
  },
  "metrics": {
    "append_only_fixture_negative_control_character_growth": {
      "mean": 12919,
      "std": 112.58774356030055,
      "n": 3
    },
    "discarded_reasoning_replayed": {
      "mean": 0,
      "std": 0.0,
      "n": 3
    },
    "executed_inert_actions": {
      "mean": 200,
      "std": 0.0,
      "n": 3
    },
    "invalid_patch_retries": {
      "mean": 1,
      "std": 0.0,
      "n": 3
    },
    "max_prompt_characters": {
      "mean": 121.66666666666667,
      "std": 0.5773502691896257,
      "n": 3
    },
    "prompt_character_range": {
      "mean": 2.6666666666666665,
      "std": 0.5773502691896257,
      "n": 3
    },
    "public_turns": {
      "mean": 200,
      "std": 0.0,
      "n": 3
    },
    "recursive_deletion_reference_error": {
      "mean": 0,
      "std": 0.0,
      "n": 3
    },
    "retry_public_context_drift": {
      "mean": 0,
      "std": 0.0,
      "n": 3
    },
    "state_update_reference_max_error": {
      "mean": 0,
      "std": 0.0,
      "n": 3
    }
  },
  "per_seed": [
    {
      "seed": 42,
      "fixture_sha256": "f45792bcb4c64c69cb7fad24e279948b8a8e9bfcb3a03479d5d897587ff5ff49",
      "metrics": {
        "public_turns": 200,
        "executed_inert_actions": 200,
        "invalid_patch_retries": 1,
        "state_update_reference_max_error": 0,
        "recursive_deletion_reference_error": 0,
        "discarded_reasoning_replayed": 0,
        "retry_public_context_drift": 0,
        "max_prompt_characters": 122,
        "prompt_character_range": 3,
        "append_only_fixture_negative_control_character_growth": 12985
      }
    },
    {
      "seed": 43,
      "fixture_sha256": "6fce50bb81c75a9aa813af888799aa7e6d585e121e2a4cf5c626160c60e00bc5",
      "metrics": {
        "public_turns": 200,
        "executed_inert_actions": 200,
        "invalid_patch_retries": 1,
        "state_update_reference_max_error": 0,
        "recursive_deletion_reference_error": 0,
        "discarded_reasoning_replayed": 0,
        "retry_public_context_drift": 0,
        "max_prompt_characters": 122,
        "prompt_character_range": 3,
        "append_only_fixture_negative_control_character_growth": 12789
      }
    },
    {
      "seed": 44,
      "fixture_sha256": "927ccc0549ab23c67efa77377e8fdf90cf5e6f7a7d197551e412bb134700cc63",
      "metrics": {
        "public_turns": 200,
        "executed_inert_actions": 200,
        "invalid_patch_retries": 1,
        "state_update_reference_max_error": 0,
        "recursive_deletion_reference_error": 0,
        "discarded_reasoning_replayed": 0,
        "retry_public_context_drift": 0,
        "max_prompt_characters": 121,
        "prompt_character_range": 2,
        "append_only_fixture_negative_control_character_growth": 12983
      }
    }
  ],
  "provenance": {
    "commit": "8c963f810e45ce3886e43b1b5da39980e41c95c0",
    "artifact_path": "docs/agent-research/2608.26263-skill-state/metrics/component-cpu-seeds42-44.json",
    "dataset_fingerprint": "sha256:7d3e31b265901c7e970d4289271f67e9fcbe3032d63a7a50469b1c9b448b410c",
    "raw_runner_artifact_sha256": "760064cde597a192be60ab2076576af1a97bb5994243e1ea4d81e703f53e1d92",
    "executed_source_sha256": {
      "scripts/run_oct19_skill_state.py": "390834ddc627a1c75cbc49d9efd8202bec765e023bf0c30dd642cd6c8ba473b0",
      "src/auto_research/agent_research/skill_state.py": "21d1f12a09c1865c3e367f9246d94c93c19c2d6599f5d96b76a8b7db51cec6b1"
    },
    "accelerator": "CPU"
  },
  "diagnostic_only": true,
  "promotion_eligible": false,
  "evolve_integrated": false,
  "full_experiment_status": "not_reproduced_high_cost",
  "full_experiment_review_done": true
}
