{
  "schema_version": 1,
  "suite_version": "agentic-v2.0.1",
  "suite_fingerprint": "e6561c8c2a4f8337e5d7fec02ee3eff01f9eafe1157b601af2fb79fec4e316c0",
  "tasks": {
    "cache": {
      "language": "python",
      "turns": 1,
      "summary": "Expiration and LRU without API regressions"
    },
    "ledger": {
      "language": "python",
      "turns": 1,
      "summary": "Exact money, CSV and failure-safe CLI"
    },
    "buildplan": {
      "language": "python",
      "turns": 1,
      "summary": "Deterministic graph planning and compatibility"
    },
    "wallet": {
      "language": "python",
      "turns": 1,
      "summary": "Atomic and concurrent transfers with overflow checks"
    },
    "async_pool": {
      "language": "typescript",
      "turns": 1,
      "summary": "Bounded async scheduling and failure recovery"
    },
    "pagination": {
      "language": "typescript",
      "turns": 1,
      "summary": "Multi-file pagination, deduplication and cancellation"
    },
    "eventstream": {
      "language": "python",
      "turns": 1,
      "summary": "Incremental UTF-8 parsing across arbitrary chunk boundaries"
    },
    "boltons_paths": {
      "language": "python",
      "turns": 1,
      "summary": "Locate a nested-lookup regression in an upstream library"
    }
  },
  "profiles": {
    "medium_8k": {
      "context_tokens": 131072,
      "output_tokens_per_response": 8192,
      "enable_thinking": true,
      "reasoning_effort": "medium",
      "temperature": 1.0,
      "temperature_origin": "server_default",
      "top_p": 0.95,
      "top_k": 20,
      "min_p": 0.0,
      "repeat_penalty": 1.0
    },
    "medium_16k": {
      "context_tokens": 131072,
      "output_tokens_per_response": 16384,
      "enable_thinking": true,
      "reasoning_effort": "medium",
      "temperature": 1.0,
      "temperature_origin": "server_default",
      "top_p": 0.95,
      "top_k": 20,
      "min_p": 0.0,
      "repeat_penalty": 1.0
    },
    "medium_32k": {
      "context_tokens": 131072,
      "output_tokens_per_response": 32768,
      "enable_thinking": true,
      "reasoning_effort": "medium",
      "temperature": 1.0,
      "temperature_origin": "server_default",
      "top_p": 0.95,
      "top_k": 20,
      "min_p": 0.0,
      "repeat_penalty": 1.0
    },
    "xhigh_32k": {
      "context_tokens": 131072,
      "output_tokens_per_response": 32768,
      "enable_thinking": true,
      "reasoning_effort": "xhigh",
      "temperature": 1.0,
      "temperature_origin": "server_default",
      "top_p": 0.95,
      "top_k": 20,
      "min_p": 0.0,
      "repeat_penalty": 1.0
    },
    "xhigh_32k_temperature08": {
      "context_tokens": 131072,
      "output_tokens_per_response": 32768,
      "enable_thinking": true,
      "reasoning_effort": "xhigh",
      "temperature": 0.8,
      "temperature_origin": "profile_override",
      "top_p": 0.95,
      "top_k": 20,
      "min_p": 0.0,
      "repeat_penalty": 1.0
    },
    "thinking_off_32k": {
      "context_tokens": 131072,
      "output_tokens_per_response": 32768,
      "enable_thinking": false,
      "reasoning_effort": "medium",
      "temperature": 1.0,
      "temperature_origin": "server_default",
      "top_p": 0.95,
      "top_k": 20,
      "min_p": 0.0,
      "repeat_penalty": 1.0
    }
  },
  "quality_budget": {
    "seconds": 1800,
    "generated_tokens": 65536,
    "tool_calls": 120
  },
  "cohorts": {
    "medium_8k": {
      "attempt_ids": [
        "a0001",
        "a0002",
        "a0003",
        "a0004",
        "a0005",
        "a0006",
        "a0007",
        "a0008",
        "a0009",
        "a0010",
        "a0011",
        "a0012",
        "a0013",
        "a0014",
        "a0015",
        "a0016",
        "a0017",
        "a0018",
        "a0019",
        "a0020",
        "a0021",
        "a0022",
        "a0023",
        "a0024"
      ],
      "expected_count": 24,
      "role": "primary_development"
    },
    "medium_16k": {
      "attempt_ids": [
        "a0025",
        "a0026",
        "a0027",
        "a0028",
        "a0029",
        "a0030",
        "a0031",
        "a0032",
        "a0033",
        "a0034",
        "a0035",
        "a0036",
        "a0037",
        "a0038",
        "a0039",
        "a0040",
        "a0041",
        "a0042",
        "a0043",
        "a0044",
        "a0045",
        "a0046",
        "a0047",
        "a0048"
      ],
      "expected_count": 24,
      "role": "primary_development"
    },
    "medium_32k": {
      "attempt_ids": [
        "a0049",
        "a0050",
        "a0051",
        "a0052",
        "a0053",
        "a0054",
        "a0055",
        "a0056",
        "a0057",
        "a0058",
        "a0059",
        "a0060",
        "a0061",
        "a0062",
        "a0063",
        "a0064",
        "a0065",
        "a0066",
        "a0067",
        "a0068",
        "a0069",
        "a0070",
        "a0071",
        "a0072"
      ],
      "expected_count": 24,
      "role": "primary_development"
    },
    "xhigh_32k": {
      "attempt_ids": [
        "a0073",
        "a0074",
        "a0075",
        "a0076",
        "a0077",
        "a0078",
        "a0079",
        "a0080",
        "a0081",
        "a0082",
        "a0083",
        "a0084",
        "a0085",
        "a0086",
        "a0087",
        "a0088",
        "a0089",
        "a0090",
        "a0091",
        "a0092",
        "a0093",
        "a0094",
        "a0095",
        "a0096"
      ],
      "expected_count": 24,
      "role": "primary_development"
    },
    "temperature08_targeted": {
      "attempt_ids": [
        "a0097",
        "a0098",
        "a0099",
        "a0100",
        "a0101",
        "a0102"
      ],
      "expected_count": 6,
      "role": "targeted_development"
    },
    "temperature10_reused_controls": {
      "attempt_ids": [
        "a0078",
        "a0079",
        "a0081",
        "a0084",
        "a0089",
        "a0092"
      ],
      "expected_count": 6,
      "role": "reused_subset_of_xhigh_32k"
    }
  },
  "comparisons": [
    [
      "medium_8k",
      "medium_16k"
    ],
    [
      "medium_16k",
      "medium_32k"
    ],
    [
      "medium_32k",
      "xhigh_32k"
    ],
    [
      "temperature10_reused_controls",
      "temperature08_targeted"
    ]
  ],
  "expected_inventory": {
    "unique_records": 159,
    "completed_results": 119,
    "primary_unique_attempts": 102,
    "extra_8k_excluded": 14,
    "pilots": 2,
    "thinking_off_completed": 1,
    "thinking_off_cancelled_after_start": 1,
    "thinking_off_cancelled_unstarted": 22,
    "infrastructure_no_result": 3,
    "configuration_only": 4,
    "extra_8k_cancelled_unstarted": 10
  },
  "repeat_policy": "Date-shaped seeds omitted. repeat-1, repeat-2 and repeat-3 consistently preserve the original paired seed order; pilot-repeat-1 is separate. Labels are not numeric inference seeds.",
  "temperature_sampling": [
    {
      "attempt_id": "a0097",
      "observations": 25,
      "all_sampled_temperatures_match_08": true
    },
    {
      "attempt_id": "a0098",
      "observations": 11,
      "all_sampled_temperatures_match_08": true
    },
    {
      "attempt_id": "a0099",
      "observations": 10,
      "all_sampled_temperatures_match_08": true
    },
    {
      "attempt_id": "a0100",
      "observations": 17,
      "all_sampled_temperatures_match_08": true
    },
    {
      "attempt_id": "a0101",
      "observations": 16,
      "all_sampled_temperatures_match_08": true
    },
    {
      "attempt_id": "a0102",
      "observations": 16,
      "all_sampled_temperatures_match_08": true
    }
  ],
  "replacement_attempt_id": "a0038",
  "technical_identity": {
    "model_filename": "Qwen3.8-27B-Q4_K_M.gguf",
    "weights_sha256": "f5f1dd8920d417aac2718b0bda3403da274301efdd6760b4f0f4b864ff2ad57d",
    "weights_bytes": 16810714464,
    "runtime_build": "b11146-7fe450e19",
    "opencode_sha256": "a5f6b8fa55c21b4897bf640bbedfa64c06c11aa17731b6f81e9ea756dcf94813",
    "versions": {
      "python": "Python 3.12.3",
      "node": "v24.21.0",
      "typescript": "Version 5.9.3",
      "opencode": "opencode v2.0.20"
    },
    "identity_basis": "Historical per-attempt recorded full-weight verification, configuration and server properties; no fresh weight hashing or live service inspection. Recorded deployment name is not independent upstream authenticity verification.",
    "hardware": "NVIDIA GeForce RTX 3090, 24 GiB (accepted historical audit)",
    "cuda_version": "12.8 (accepted historical deployment audit)",
    "serving_settings": {
      "gpu_layers": "all",
      "fit": "off",
      "flash_attention": "on",
      "cache_k": "q8_0",
      "cache_v": "q8_0",
      "batch_size": 512,
      "microbatch_size": 256,
      "context_checkpoints": 4,
      "host_cache_mib": 2048,
      "threads": 8,
      "threads_batch": 8,
      "parallel_slots": 1,
      "context_shift": false,
      "vision_projector": false,
      "speculation": "none",
      "reasoning_format": "deepseek",
      "reasoning_preserve": true
    }
  },
  "metric_definitions": {
    "functional_success": "Solved phase-one AND final grade AND unchanged configuration.",
    "grade_solved": "No protected changes AND no missing starter files AND every required check group successful.",
    "delivered_success": "Functional success AND expected turn count AND every turn completed with exit zero, nonempty final text after tools, last recognized test validation passed, and last recognized typecheck passed for TypeScript.",
    "seconds": "Sum of recorded turn durations; excludes startup, between-attempt rests and independent grading.",
    "native_output_tokens": "Sum of model-server generated-token counter deltas over turns.",
    "opencode_output_tokens": "Sum of output usage reported by OpenCode over turns.",
    "missing_result": "Null outcome, never a scored model failure.",
    "paired_transition": "Match by case and public repeat label; compare stored and recomputed booleans.",
    "median_seconds": "Median of all completed attempts in the selected cohort, including failures."
  },
  "lineage": {
    "development_task_count": 8,
    "turns_per_task": 1,
    "boltons_paths_upstream_revision": "4e5faa3d7e4008d89e0d8bf1ea87b6d9a061a16d",
    "boltons_paths_license": "BSD-3-Clause",
    "boltons_paths_kind": "pinned_upstream_seeded_fault",
    "other_tasks_kind": "authored_fixture"
  }
}