{
  "schema_version": "2026-05-24",
  "suite_id": "codex-guided",
  "source": "scripts/benchmark-repeat.sh aggregate.json",
  "reproducibility": {
    "manifest_sha256": "63b4c20eb58082ceb35dcabdaa43ac21c1c187313f4498bafa57aa160bdb81e9",
    "manifest_schema_version": "2026-05-24",
    "harness": "codex",
    "required_repeats": 3,
    "fresh_context_contract": "Each repeat invokes one full baseline/optimized harness run. The harness creates fresh detached git worktrees and launches separate Claude/Codex task sessions for baseline and optimized runs.",
    "base_ref": "b96b8a7f5cc57c4335bc7bc85ec726c836ed0996",
    "quality_command": "go test ./..."
  },
  "required_repeats": 3,
  "completed_repeats": 3,
  "quality_passed": true,
  "failed_repeats": [],
  "delta_stats": {
    "avoidable_waste_high_points": {
      "n": 3,
      "mean": -1.3333333333333333,
      "median": 0,
      "min": -6,
      "max": 2,
      "stdev": 4.163331998932265
    },
    "avoidable_waste_low_points": {
      "n": 3,
      "mean": -1.3333333333333333,
      "median": 0,
      "min": -6,
      "max": 2,
      "stdev": 4.163331998932265
    },
    "codex_cached_input_tokens": {
      "n": 3,
      "mean": -75434.66666666667,
      "median": -69632,
      "min": -148224,
      "max": -8448,
      "stdev": 70068.4361844428
    },
    "codex_input_tokens": {
      "n": 3,
      "mean": -99320.0,
      "median": -80685,
      "min": -190460,
      "max": -26815,
      "stdev": 83398.85445855957
    },
    "codex_output_tokens": {
      "n": 3,
      "mean": -483.3333333333333,
      "median": -571,
      "min": -654,
      "max": -225,
      "stdev": 227.53974011880504
    },
    "codex_reasoning_output_tokens": {
      "n": 3,
      "mean": -44.666666666666664,
      "median": -22,
      "min": -96,
      "max": -16,
      "stdev": 44.55707949735186
    },
    "codex_total_uncached_plus_output_tokens": {
      "n": 3,
      "mean": -24368.666666666668,
      "median": -19021,
      "min": -42807,
      "max": -11278,
      "stdev": 16430.691231148292
    },
    "context_growth_events": {
      "n": 3,
      "mean": 0.0,
      "median": 0,
      "min": 0,
      "max": 0,
      "stdev": 0.0
    },
    "estimated_tokens": {
      "n": 3,
      "mean": -14520.333333333334,
      "median": -15547,
      "min": -18399,
      "max": -9615,
      "stdev": 4481.093318971759
    },
    "failed_commands": {
      "n": 3,
      "mean": 0.3333333333333333,
      "median": 0,
      "min": 0,
      "max": 1,
      "stdev": 0.5773502691896257
    },
    "rereads": {
      "n": 3,
      "mean": -1.0,
      "median": 0,
      "min": -3,
      "max": 0,
      "stdev": 1.7320508075688772
    },
    "tool_output_tokens": {
      "n": 3,
      "mean": -14526.666666666666,
      "median": -15541,
      "min": -18416,
      "max": -9623,
      "stdev": 4483.3989710189
    }
  },
  "run_dirs": [
    "run-01",
    "run-02",
    "run-03"
  ],
  "promotion_policy": "recommendation_evidence",
  "candidate_source_url": null,
  "candidate_issue": null,
  "published_api_cost_estimate": {
    "harness": "codex",
    "complete_cost_surface": true,
    "multi_session_runs_not_in_stdout_usage": [],
    "rates": {
      "model_assumption": "GPT-5.3-Codex Standard",
      "input_tokens_per_mtok_usd": 1.75,
      "cached_input_tokens_per_mtok_usd": 0.175,
      "output_tokens_per_mtok_usd": 14.0,
      "reasoning_output_tokens_per_mtok_usd": 14.0
    },
    "baseline_mean_usd": 0.19605973333333335,
    "optimized_mean_usd": 0.13366733333333333,
    "delta_mean_usd": -0.062392399999999994,
    "savings_rate": 0.31823158656409467,
    "savings_percent": 31.82315865640947,
    "scaled_examples": {
      "savings_rate": 0.31823158656409467,
      "savings_percent": 31.82315865640947,
      "weekly_100_usd": 31.82315865640947,
      "weekly_500_usd": 159.11579328204735,
      "monthly_2000_usd": 636.4631731281894,
      "monthly_5000_usd": 1591.1579328204734,
      "monthly_10000_usd": 3182.315865640947,
      "basis": "Scaled linearly from the suite's optimized-vs-baseline published API-rate delta: savings = comparable baseline spend * savings_percent."
    },
    "delta_min_usd": -0.1080702,
    "delta_max_usd": -0.03602235000000001,
    "runs": [
      {
        "run": "run-01",
        "baseline_usd": 0.179473,
        "optimized_usd": 0.13638835000000002,
        "delta_usd": -0.043084649999999974
      },
      {
        "run": "run-02",
        "baseline_usd": 0.228592,
        "optimized_usd": 0.12052179999999998,
        "delta_usd": -0.1080702
      },
      {
        "run": "run-03",
        "baseline_usd": 0.1801142,
        "optimized_usd": 0.14409185,
        "delta_usd": -0.03602235000000001
      }
    ],
    "note": "Uses Codex JSON usage fields repriced at published GPT-5.3-Codex Standard API rates. Reasoning tokens are billed at the output-token rate in this estimate."
  },
  "privacy_boundary": "Raw Claude/Codex logs, raw prompts, local paths, and secrets are not included in public aggregate artifacts."
}