{
  "schema_version": 1,
  "trial": {
    "id": "csv-export-example",
    "product": "Demo Export CLI",
    "repetitions": 1,
    "parallel": 1,
    "order_seed": 0
  },
  "plan": {
    "wall_clock_comparable": true,
    "note": "Synthetic, sequential examples."
  },
  "consistency": {
    "prompt_sha256": "b10b4173c2cc01852b1db682fbfbb91ce20ceedc033cf0633f7686e38630ee76",
    "product_versions": [
      "1.2.0",
      "1.3.0"
    ],
    "harness_versions": [
      "codex synthetic fixture",
      "kiro-cli synthetic fixture"
    ],
    "wall_clock_comparable": true,
    "token_note": "output tokens use each model's tokenizer; compare across models as resource usage, not reasoning; per-cell token variance includes only runs with complete (reconciled) usage",
    "cost_note": "cost_estimate_usd (Claude Code) and credits (Kiro) are harness estimates in different units; never summed or sorted across harnesses"
  },
  "runs": [
    {
      "run_id": "baseline-codex-r1",
      "dir": "baseline-codex-r1",
      "harness": "codex",
      "harness_version": "synthetic fixture",
      "model_requested": "example-model",
      "model_id": "example-model",
      "config": "clean",
      "auth": "synthetic; no credentials",
      "product_version": "1.2.0",
      "declared": "I exported the records and verified that the CSV contains 100 data rows.",
      "verified": "succeeded",
      "stop_reason": "completed",
      "elapsed_seconds": 108.0,
      "tool_calls": 5,
      "tool_status": "partial",
      "failed_tool_calls": 1,
      "retries": 1,
      "output_tokens": 2862,
      "usage_status": "partial",
      "cost_estimate_usd": null,
      "credits": null,
      "asks": 3,
      "gates": 0,
      "ask_register": [
        {
          "title": "Make the quick-start export command executable",
          "requested_behavior": "Use the supported --output flag in the quick start, with a copyable CSV example.",
          "evidence_status": "observed_friction",
          "how_observed": "The documented --out flag failed. Command help showed --output, which worked.",
          "consequence": "The first attempt failed and required a help lookup before retrying.",
          "event_refs": [
            "E-003",
            "E-004",
            "E-005"
          ],
          "journey_anchors": [],
          "labels": [
            "Roadblock",
            "Misdirection"
          ],
          "priority_rationale": "Fix the broken first attempt before reducing optional friction.",
          "verification": "Run the documented command in a fresh workspace; it must produce a 100-row CSV.",
          "workaround": "Replace --out with --output.",
          "modeled_saving": null,
          "id": "ASK-001",
          "invalid_refs": [],
          "measured": {
            "events": 3,
            "tool_calls": 3,
            "failed_tool_calls": 1,
            "tool_seconds": 51.0,
            "span_seconds": 69.0,
            "output_tokens": 813,
            "shared_events": [
              "E-005"
            ],
            "basis": "sum over referenced events; tokens attributed per whole assistant message (two asks citing events from the same message both carry its tokens); span = first referenced event start to last referenced event end"
          }
        },
        {
          "title": "Expose progress while a CSV export is running",
          "requested_behavior": "Emit bounded progress on stderr and a final structured completion result.",
          "evidence_status": "observed_friction",
          "how_observed": "Only start and completion messages appeared during the 47-second export.",
          "consequence": "The agent could not distinguish productive work from a stalled command.",
          "event_refs": [
            "E-005",
            "E-006"
          ],
          "journey_anchors": [],
          "labels": [
            "Wait",
            "Blind Spot"
          ],
          "priority_rationale": "Make long-running work observable without extra polling.",
          "verification": "Run a delayed export; progress must remain separate from CSV stdout.",
          "workaround": null,
          "modeled_saving": null,
          "id": "ASK-002",
          "invalid_refs": [],
          "measured": {
            "events": 2,
            "tool_calls": 1,
            "failed_tool_calls": 0,
            "tool_seconds": 47.0,
            "span_seconds": 47.0,
            "output_tokens": 683,
            "shared_events": [
              "E-005"
            ],
            "basis": "sum over referenced events; tokens attributed per whole assistant message (two asks citing events from the same message both carry its tokens); span = first referenced event start to last referenced event end"
          }
        },
        {
          "title": "Explain when to choose export instead of dump",
          "requested_behavior": "State the output formats and intended use of both commands in top-level help.",
          "evidence_status": "observed_friction",
          "how_observed": "Both commands were described as writing records; the required CSV path was unclear.",
          "consequence": "The agent had to choose without a selection signal.",
          "event_refs": [
            "E-001",
            "E-002"
          ],
          "journey_anchors": [],
          "labels": [
            "Fork"
          ],
          "priority_rationale": "Remove the ambiguous choice at the start of the task.",
          "verification": "A user reading only top-level help can identify the CSV command.",
          "workaround": null,
          "modeled_saving": null,
          "id": "ASK-003",
          "invalid_refs": [],
          "measured": {
            "events": 2,
            "tool_calls": 1,
            "failed_tool_calls": 0,
            "tool_seconds": 2.0,
            "span_seconds": 8.0,
            "output_tokens": 307,
            "shared_events": [],
            "basis": "sum over referenced events; tokens attributed per whole assistant message (two asks citing events from the same message both carry its tokens); span = first referenced event start to last referenced event end"
          }
        }
      ],
      "no_obstacles_observed": false,
      "defects": 0,
      "journey_provenance": "reconstruction",
      "extraction_status": "ok",
      "review_status": "ready",
      "stages": {
        "execute": "done",
        "verify": "done",
        "narrate": "done",
        "extract": "done",
        "measure": "done",
        "render": "done"
      },
      "errors": [],
      "teardown": "none defined",
      "verify_auth_errors": 0,
      "limitations": 3
    },
    {
      "run_id": "headless-kiro-cli-r1",
      "dir": "headless-kiro-cli-r1",
      "harness": "kiro-cli",
      "harness_version": "synthetic fixture",
      "model_requested": "example-model",
      "model_id": "example-model",
      "config": "clean",
      "auth": "synthetic; no credentials",
      "product_version": "1.3.0",
      "declared": "The export is blocked by required approval. I did not create the CSV.",
      "verified": "failed",
      "stop_reason": "completed",
      "elapsed_seconds": 25.0,
      "tool_calls": 2,
      "tool_status": "partial",
      "failed_tool_calls": 1,
      "retries": 0,
      "output_tokens": null,
      "usage_status": "unavailable",
      "cost_estimate_usd": null,
      "credits": null,
      "asks": 1,
      "gates": 1,
      "ask_register": [
        {
          "title": "Check headless authorization before starting an export",
          "requested_behavior": "Provide a non-interactive authorization preflight and document the required human approval.",
          "evidence_status": "observed_friction",
          "how_observed": "The command requested browser approval after the agent had started the task.",
          "consequence": "The unattended task stopped without an output file.",
          "event_refs": [
            "E-002",
            "E-003",
            "E-004"
          ],
          "journey_anchors": [],
          "labels": [
            "Gate",
            "Wall"
          ],
          "priority_rationale": "Expose the prerequisite early so teams can prepare an authorized run.",
          "verification": "In headless mode, preflight reports the missing approval before an export begins.",
          "workaround": null,
          "modeled_saving": null,
          "id": "ASK-001",
          "invalid_refs": [],
          "measured": {
            "events": 3,
            "tool_calls": 1,
            "failed_tool_calls": 1,
            "tool_seconds": 3.0,
            "span_seconds": 16.0,
            "output_tokens": null,
            "shared_events": [],
            "basis": "sum over referenced events; tokens attributed per whole assistant message (two asks citing events from the same message both carry its tokens); span = first referenced event start to last referenced event end"
          }
        }
      ],
      "no_obstacles_observed": false,
      "defects": 0,
      "journey_provenance": "reconstruction",
      "extraction_status": "ok",
      "review_status": "ready",
      "stages": {
        "execute": "done",
        "verify": "done",
        "narrate": "done",
        "extract": "done",
        "measure": "done",
        "render": "done"
      },
      "errors": [],
      "teardown": "none defined",
      "verify_auth_errors": 0,
      "limitations": 3
    },
    {
      "run_id": "revised-codex-r1",
      "dir": "revised-codex-r1",
      "harness": "codex",
      "harness_version": "synthetic fixture",
      "model_requested": "example-model",
      "model_id": "example-model",
      "config": "clean",
      "auth": "synthetic; no credentials",
      "product_version": "1.3.0",
      "declared": "I created the CSV and verified all 100 data rows.",
      "verified": "succeeded",
      "stop_reason": "completed",
      "elapsed_seconds": 34.0,
      "tool_calls": 3,
      "tool_status": "partial",
      "failed_tool_calls": 0,
      "retries": 0,
      "output_tokens": 1120,
      "usage_status": "partial",
      "cost_estimate_usd": null,
      "credits": null,
      "asks": 1,
      "gates": 0,
      "ask_register": [
        {
          "title": "Expose progress while a CSV export is running",
          "requested_behavior": "Emit bounded progress on stderr and a final structured completion result.",
          "evidence_status": "observed_friction",
          "how_observed": "Only start and completion messages appeared during the 20-second export.",
          "consequence": "The agent could not observe intermediate progress.",
          "event_refs": [
            "E-002",
            "E-003"
          ],
          "journey_anchors": [],
          "labels": [
            "Wait"
          ],
          "priority_rationale": "The remaining observed friction is visibility during execution.",
          "verification": "Run a delayed export; progress must remain separate from CSV stdout.",
          "workaround": null,
          "modeled_saving": null,
          "id": "ASK-001",
          "invalid_refs": [],
          "measured": {
            "events": 2,
            "tool_calls": 1,
            "failed_tool_calls": 0,
            "tool_seconds": 20.0,
            "span_seconds": 20.0,
            "output_tokens": 401,
            "shared_events": [],
            "basis": "sum over referenced events; tokens attributed per whole assistant message (two asks citing events from the same message both carry its tokens); span = first referenced event start to last referenced event end"
          }
        }
      ],
      "no_obstacles_observed": false,
      "defects": 0,
      "journey_provenance": "reconstruction",
      "extraction_status": "ok",
      "review_status": "ready",
      "stages": {
        "execute": "done",
        "verify": "done",
        "narrate": "done",
        "extract": "done",
        "measure": "done",
        "render": "done"
      },
      "errors": [],
      "teardown": "none defined",
      "verify_auth_errors": 0,
      "limitations": 3
    }
  ],
  "cells": [
    {
      "cell": "baseline-codex",
      "runs": 1,
      "harness": "codex",
      "model_requested": "example-model",
      "model_ids": [
        "example-model"
      ],
      "config": "clean",
      "verified": [
        "succeeded"
      ],
      "elapsed_seconds": {
        "n": 1,
        "min": 108.0,
        "median": 108.0,
        "max": 108.0
      },
      "tool_calls": {
        "n": 1,
        "min": 5,
        "median": 5,
        "max": 5
      },
      "failed_tool_calls": {
        "n": 1,
        "min": 1,
        "median": 1,
        "max": 1
      },
      "output_tokens": null,
      "asks": {
        "n": 1,
        "min": 3,
        "median": 3,
        "max": 3
      }
    },
    {
      "cell": "headless-kiro-cli",
      "runs": 1,
      "harness": "kiro-cli",
      "model_requested": "example-model",
      "model_ids": [
        "example-model"
      ],
      "config": "clean",
      "verified": [
        "failed"
      ],
      "elapsed_seconds": {
        "n": 1,
        "min": 25.0,
        "median": 25.0,
        "max": 25.0
      },
      "tool_calls": {
        "n": 1,
        "min": 2,
        "median": 2,
        "max": 2
      },
      "failed_tool_calls": {
        "n": 1,
        "min": 1,
        "median": 1,
        "max": 1
      },
      "output_tokens": null,
      "asks": {
        "n": 1,
        "min": 1,
        "median": 1,
        "max": 1
      }
    },
    {
      "cell": "revised-codex",
      "runs": 1,
      "harness": "codex",
      "model_requested": "example-model",
      "model_ids": [
        "example-model"
      ],
      "config": "clean",
      "verified": [
        "succeeded"
      ],
      "elapsed_seconds": {
        "n": 1,
        "min": 34.0,
        "median": 34.0,
        "max": 34.0
      },
      "tool_calls": {
        "n": 1,
        "min": 3,
        "median": 3,
        "max": 3
      },
      "failed_tool_calls": {
        "n": 1,
        "min": 0,
        "median": 0,
        "max": 0
      },
      "output_tokens": null,
      "asks": {
        "n": 1,
        "min": 1,
        "median": 1,
        "max": 1
      }
    }
  ],
  "clusters": null,
  "generated_at": "2026-10-07T10:28:00.238Z"
}