okstra 0.102.2 → 0.102.3

package/docs/superpowers/plans/2026-06-20-run-index-row-normalization.md

@@ -1,583 +0,0 @@
-# run-index row 정규화 (runId SSOT화) Implementation Plan
-
-> **For agentic workers:** REQUIRED SUB-SKILL: Use superpowers:subagent-driven-development (recommended) or superpowers:executing-plans to implement this plan task-by-task. Steps use checkbox (`- [ ]`) syntax for tracking.
-
-**Goal:** run-index row(`recent.jsonl`/`active.jsonl`/`projects/<id>/index.jsonl`)에서 runId로부터 무손실 복원 가능한 5개 파생 필드를 제거해 디스크 포맷을 slim(18→13필드)으로 만들고, read 지점 hydrate로 reader 로직을 무변경 유지한다.
-
-**Architecture:** 신규 모듈 `run_index_row.py`에 생성(`build_run_index_row`)·축약(`slim_run_row`)·복원(`hydrate_run_row`)·읽기(`read_run_index`)·일괄 변환(`reindex_slim`)을 집중한다. writer는 `build_run_index_row`로 slim row만 기록하고, reader 3곳은 `read_run_index`로 읽어 full row를 받는다. reconcile은 읽을 때 hydrate, 다시 쓸 때 slim한다.
-
-**Tech Stack:** Python 3 (4-space indent, snake_case), pytest (`tests/run/`, conftest.py가 OKSTRA_HOME 격리), `npm run build`로 runtime 동기화.
-
-## Global Constraints
-
-- 디스크 row에서 제거하는 필드(정확히 5개): `projectId`, `taskType`, `runSeq`, `taskKey`, `invocationFile`.
-- 유지하는 필드(13개): `runId`, `projectRoot`, `taskGroup`, `taskId`, `status`, `startedAt`, `finishedAt`, `workers`, `leadModel`, `validation`, `finalReportRel`, `finalStatusRel`, `runDirRel`.
-- `taskGroup`/`taskId`는 raw 대소문자를 보존한다 (runId 세그먼트는 slug라 다를 수 있음 — 예: row `DEV-9388` vs runId `dev-9388`). 절대 runId에서 복원하지 않는다.
-- `runId` 포맷 `<projectId>/<taskGroup>/<taskId>/<taskType>/r<NN>`은 fs-safe 슬러그 불변식이라 변경 금지.
-- 모든 row 키는 camelCase. JSON 직렬화는 기존과 동일하게 `separators=(",", ":")`.
-- `runtime/`는 빌드 산출물 — 직접 편집 금지. 소스 수정 후 `npm run build`.
-- 커밋은 Conventional Commits, AI 트레일러 금지.
-
----
-
-### Task 1: 코어 헬퍼 — build / slim / hydrate
-
-**Files:**
-- Create: `scripts/okstra_ctl/run_index_row.py`
-- Test: `tests/run/test_run_index_row.py`
-
-**Interfaces:**
-- Consumes: `parse_run_id`, `build_run_id` from [`scripts/okstra_ctl/ids.py`](../../../scripts/okstra_ctl/ids.py); `invocation_path` from [`scripts/okstra_ctl/invocation.py`](../../../scripts/okstra_ctl/invocation.py).
-- Produces:
-  - `DERIVED_FIELDS: tuple = ("projectId", "taskType", "runSeq", "taskKey", "invocationFile")`
-  - `build_run_index_row(*, project_id, project_root, task_group, task_id, task_type, run_seq, status, started_at, finished_at, workers, lead_model, validation, run_dir_rel, final_report_rel, final_status_rel) -> dict` — 13필드 slim row 반환.
-  - `slim_run_row(row: dict) -> dict` — `DERIVED_FIELDS`를 제거한 새 dict (멱등).
-  - `hydrate_run_row(row: dict) -> dict` — slim/full row를 받아 5개 파생 필드를 채운 full dict (멱등).
-
-- [ ] **Step 1: Write the failing test**
-
-```python
-# tests/run/test_run_index_row.py
-from pathlib import Path
-
-from okstra_ctl.run_index_row import (
-    DERIVED_FIELDS, build_run_index_row, slim_run_row, hydrate_run_row,
-)
-
-
-def _sample_kwargs():
-    return dict(
-        project_id="fontradar-v2-api",
-        project_root="/Volumes/x/fontradar-v2-api",
-        task_group="uploadfont",
-        task_id="dev-6827-bug-de-redis-lupload",
-        task_type="final-verification",
-        run_seq=5,
-        status="prepared",
-        started_at="2026-06-18T20:24:00Z",
-        finished_at="2026-06-18T20:24:00Z",
-        workers=["claude", "codex"],
-        lead_model="opus",
-        validation="not-run",
-        run_dir_rel=".okstra/tasks/uploadfont/dev-6827-bug-de-redis-lupload/runs/final-verification",
-        final_report_rel=".okstra/tasks/uploadfont/dev-6827-bug-de-redis-lupload/runs/final-verification/reports/final-report-final-verification-001.md",
-        final_status_rel=".okstra/tasks/uploadfont/dev-6827-bug-de-redis-lupload/runs/final-verification/status/final-final-verification-001.status",
-    )
-
-
-def test_build_run_index_row_omits_derived():
-    row = build_run_index_row(**_sample_kwargs())
-    for f in DERIVED_FIELDS:
-        assert f not in row, f"slim row 에 파생 필드 {f} 가 남았다"
-    assert row["runId"] == ("fontradar-v2-api/uploadfont/"
-                            "dev-6827-bug-de-redis-lupload/final-verification/r05")
-    assert row["taskGroup"] == "uploadfont"
-    assert len(row) == 13
-
-
-def test_hydrate_restores_derived():
-    slim = build_run_index_row(**_sample_kwargs())
-    full = hydrate_run_row(slim)
-    assert full["projectId"] == "fontradar-v2-api"
-    assert full["taskType"] == "final-verification"
-    assert full["runSeq"] == 5
-    assert full["taskKey"] == "uploadfont/dev-6827-bug-de-redis-lupload"
-    assert full["invocationFile"] == (
-        "projects/fontradar-v2-api/invocations/uploadfont/"
-        "dev-6827-bug-de-redis-lupload/final-verification/r05.json")
-
-
-def test_hydrate_preserves_raw_case():
-    # row 의 raw 대소문자(DEV-9388)는 runId(slug dev-9388)에서 복원하면 안 된다
-    kwargs = _sample_kwargs()
-    kwargs["task_group"] = "DEV-9388"
-    kwargs["task_id"] = "dev-9436"
-    slim = build_run_index_row(**kwargs)
-    full = hydrate_run_row(slim)
-    assert full["taskGroup"] == "DEV-9388"
-    assert full["taskKey"] == "DEV-9388/dev-9436"
-    # invocationFile 은 slug 기반
-    assert "/dev-9388/dev-9436/" in full["invocationFile"]
-
-
-def test_hydrate_idempotent_on_full():
-    slim = build_run_index_row(**_sample_kwargs())
-    once = hydrate_run_row(slim)
-    twice = hydrate_run_row(once)
-    assert once == twice
-
-
-def test_slim_run_row_idempotent():
-    slim = build_run_index_row(**_sample_kwargs())
-    full = hydrate_run_row(slim)
-    assert slim_run_row(full) == slim
-    assert slim_run_row(slim) == slim
-```
-
-- [ ] **Step 2: Run test to verify it fails**
-
-Run: `python3 -m pytest tests/run/test_run_index_row.py -v`
-Expected: FAIL — `ModuleNotFoundError: No module named 'okstra_ctl.run_index_row'`
-
-- [ ] **Step 3: Write minimal implementation**
-
-```python
-# scripts/okstra_ctl/run_index_row.py
-"""run-index row 의 생성/축약/복원 단일 참조점.
-
-디스크에는 slim(13필드)로 저장하고, reader 는 hydrate 로 full(18필드)을 받는다.
-파생 5필드(projectId/taskType/runSeq/taskKey/invocationFile)는 runId 와 평면
-taskGroup/taskId 로부터 무손실 복원 가능하다(2026-06-20 전수 검증 근거).
-"""
-from __future__ import annotations
-
-from pathlib import Path
-
-from .ids import build_run_id, parse_run_id
-from .invocation import invocation_path
-
-DERIVED_FIELDS = ("projectId", "taskType", "runSeq", "taskKey", "invocationFile")
-
-
-def build_run_index_row(*, project_id: str, project_root: str,
-                        task_group: str, task_id: str, task_type: str,
-                        run_seq: int, status: str, started_at: str,
-                        finished_at, workers: list, lead_model: str,
-                        validation: str, run_dir_rel: str,
-                        final_report_rel: str, final_status_rel: str) -> dict:
-    """slim run-index row(13필드)를 만든다. 파생 5필드는 포함하지 않는다."""
-    return {
-        "runId": build_run_id(project_id, task_group, task_id,
-                              task_type, run_seq),
-        "projectRoot": project_root,
-        "taskGroup": task_group,
-        "taskId": task_id,
-        "status": status,
-        "startedAt": started_at,
-        "finishedAt": finished_at,
-        "workers": workers,
-        "leadModel": lead_model,
-        "validation": validation,
-        "finalReportRel": final_report_rel,
-        "finalStatusRel": final_status_rel,
-        "runDirRel": run_dir_rel,
-    }
-
-
-def slim_run_row(row: dict) -> dict:
-    """파생 5필드를 제거한 새 dict. 이미 slim 이면 동일(멱등)."""
-    return {k: v for k, v in row.items() if k not in DERIVED_FIELDS}
-
-
-def hydrate_run_row(row: dict) -> dict:
-    """slim/full row 를 받아 파생 5필드를 채운 full dict. 멱등.
-    taskKey 는 runId(slug)가 아닌 평면 raw taskGroup/taskId 로 조합한다.
-    """
-    parsed = parse_run_id(row["runId"])
-    out = dict(row)
-    out["projectId"] = parsed["projectId"]
-    out["taskType"] = parsed["taskType"]
-    out["runSeq"] = parsed["runSeq"]
-    out["taskKey"] = f"{row.get('taskGroup', '')}/{row.get('taskId', '')}"
-    out["invocationFile"] = str(invocation_path(
-        Path(""), parsed["projectId"], parsed["taskGroup"], parsed["taskId"],
-        parsed["taskType"], parsed["runSeq"]))
-    return out
-```
-
-- [ ] **Step 4: Run test to verify it passes**
-
-Run: `python3 -m pytest tests/run/test_run_index_row.py -v`
-Expected: PASS (5 tests)
-
-- [ ] **Step 5: Commit**
-
-```bash
-git add scripts/okstra_ctl/run_index_row.py tests/run/test_run_index_row.py
-git commit -m "feat(index): run-index row 생성/축약/복원 코어 헬퍼 추가"
-```
-
----
-
-### Task 2: read_run_index + reindex_slim 일괄 변환
-
-**Files:**
-- Modify: `scripts/okstra_ctl/run_index_row.py` (함수 추가)
-- Test: `tests/run/test_run_index_row.py` (테스트 추가)
-
-**Interfaces:**
-- Consumes: `read_jsonl` from [`scripts/okstra_ctl/jsonl.py`](../../../scripts/okstra_ctl/jsonl.py); `hydrate_run_row`, `slim_run_row` (Task 1).
-- Produces:
-  - `read_run_index(path: Path) -> list[dict]` — jsonl을 읽어 각 row를 `hydrate_run_row`로 통과시킨 full row 목록.
-  - `reindex_slim(home: Path) -> int` — `recent.jsonl`/`active.jsonl`/`projects/*/index.jsonl`을 slim으로 원자적 재작성. 변환한 row 총수 반환. 멱등.
-
-- [ ] **Step 1: Write the failing test**
-
-```python
-# tests/run/test_run_index_row.py (추가)
-import json
-
-from okstra_ctl.run_index_row import read_run_index, reindex_slim
-
-
-def _write_full_row(path, kwargs_overrides=None):
-    # 마이그레이션 전 full row 를 디스크에 직접 기록
-    from okstra_ctl.run_index_row import hydrate_run_row, build_run_index_row
-    kw = _sample_kwargs()
-    if kwargs_overrides:
-        kw.update(kwargs_overrides)
-    full = hydrate_run_row(build_run_index_row(**kw))
-    path.parent.mkdir(parents=True, exist_ok=True)
-    with path.open("a") as f:
-        f.write(json.dumps(full, separators=(",", ":")) + "\n")
-    return full
-
-
-def test_read_run_index_hydrates(tmp_path):
-    recent = tmp_path / "recent.jsonl"
-    # slim row 를 디스크에 기록
-    from okstra_ctl.run_index_row import build_run_index_row
-    slim = build_run_index_row(**_sample_kwargs())
-    recent.write_text(json.dumps(slim, separators=(",", ":")) + "\n")
-    rows = read_run_index(recent)
-    assert rows[0]["projectId"] == "fontradar-v2-api"
-    assert rows[0]["invocationFile"].startswith("projects/")
-
-
-def test_reindex_slim_strips_and_is_idempotent(tmp_path):
-    home = tmp_path
-    recent = home / "recent.jsonl"
-    idx = home / "projects" / "fontradar-v2-api" / "index.jsonl"
-    full_recent = _write_full_row(recent)
-    _write_full_row(idx)
-    n1 = reindex_slim(home)
-    assert n1 == 2
-    # 디스크에서 파생 필드가 사라졌는지 확인
-    on_disk = json.loads(recent.read_text().splitlines()[0])
-    for f in DERIVED_FIELDS:
-        assert f not in on_disk
-    # 값 보존: hydrate 하면 원래 full 과 동일
-    assert hydrate_run_row(on_disk) == full_recent
-    # 멱등: 두 번째 실행도 같은 결과
-    n2 = reindex_slim(home)
-    assert json.loads(recent.read_text().splitlines()[0]) == on_disk
-```
-
-- [ ] **Step 2: Run test to verify it fails**
-
-Run: `python3 -m pytest tests/run/test_run_index_row.py::test_reindex_slim_strips_and_is_idempotent -v`
-Expected: FAIL — `ImportError: cannot import name 'reindex_slim'`
-
-- [ ] **Step 3: Write minimal implementation**
-
-```python
-# scripts/okstra_ctl/run_index_row.py (추가)
-import json
-import os
-
-from .jsonl import read_jsonl
-
-
-def read_run_index(path: Path) -> list:
-    """run-index jsonl 을 읽어 각 row 를 hydrate 한 full row 목록 반환."""
-    return [hydrate_run_row(r) for r in read_jsonl(path)]
-
-
-def _rewrite_slim(path: Path) -> int:
-    rows = read_jsonl(path)
-    if not rows:
-        return 0
-    tmp = path.with_suffix(path.suffix + ".tmp")
-    with tmp.open("w") as f:
-        for r in rows:
-            f.write(json.dumps(slim_run_row(r), separators=(",", ":")) + "\n")
-    os.replace(tmp, path)
-    return len(rows)
-
-
-def reindex_slim(home: Path) -> int:
-    """기존 run-index 파일들을 slim 으로 일괄 재작성. 멱등. 변환 row 총수 반환."""
-    total = 0
-    targets = [home / "recent.jsonl", home / "active.jsonl"]
-    projects = home / "projects"
-    if projects.is_dir():
-        targets += sorted(projects.glob("*/index.jsonl"))
-    for path in targets:
-        if path.is_file():
-            total += _rewrite_slim(path)
-    return total
-```
-
-- [ ] **Step 4: Run test to verify it passes**
-
-Run: `python3 -m pytest tests/run/test_run_index_row.py -v`
-Expected: PASS (7 tests)
-
-- [ ] **Step 5: Commit**
-
-```bash
-git add scripts/okstra_ctl/run_index_row.py tests/run/test_run_index_row.py
-git commit -m "feat(index): read_run_index hydrate 와 reindex_slim 일괄 변환 추가"
-```
-
----
-
-### Task 3: writer 배선 — index.py / backfill.py
-
-**Files:**
-- Modify: `scripts/okstra_ctl/index.py:74-97` (record_start), `scripts/okstra_ctl/index.py:162-183` (reserve_run_in_active)
-- Modify: `scripts/okstra_ctl/backfill.py:219-235` (row 생성)
-- Test: 기존 `tests/run/` 회귀 (`test_okstra_index*.py`, `test_okstra_backfill*.py` 가 있으면 그 전체)
-
-**Interfaces:**
-- Consumes: `build_run_index_row` (Task 1).
-- Produces: writer 가 기록하는 모든 row 가 slim(13필드). runId 값과 인덱스 invariant는 불변.
-
-- [ ] **Step 1: Modify record_start to use build_run_index_row**
-
-[`scripts/okstra_ctl/index.py:74-97`](../../../scripts/okstra_ctl/index.py:74)의 `row = { ... }` 리터럴 전체를 다음으로 치환:
-
-```python
-    row = build_run_index_row(
-        project_id=project_id, project_root=project_root,
-        task_group=task_group, task_id=task_id, task_type=task_type,
-        run_seq=run_seq, status=initial_status, started_at=when,
-        finished_at=when if is_terminal else None,
-        workers=workers, lead_model=lead_model, validation="not-run",
-        run_dir_rel=run_dir_rel, final_report_rel=final_report_rel,
-        final_status_rel=final_status_rel)
-```
-
-파일 상단 import에 추가: `from .run_index_row import build_run_index_row`
-(`run_id`/`inv_path_abs`/`inv_path_rel` 계산 라인 67-70, save_invocation 호출은 그대로 둔다 — `row["runId"]` 대신 기존 `run_id` 변수를 계속 사용.)
-
-- [ ] **Step 2: Modify reserve_run_in_active to use build_run_index_row**
-
-[`scripts/okstra_ctl/index.py:162-183`](../../../scripts/okstra_ctl/index.py:162)의 `row = { ... }` 리터럴 전체를 치환:
-
-```python
-    row = build_run_index_row(
-        project_id=project_id, project_root=project_root,
-        task_group=task_group, task_id=task_id, task_type=task_type,
-        run_seq=run_seq, status="reserving", started_at=when,
-        finished_at=None, workers=[], lead_model="", validation="not-run",
-        run_dir_rel=run_dir_rel, final_report_rel=final_report_rel,
-        final_status_rel=final_status_rel)
-```
-
-(`inv_rel` 계산 라인 160-161은 더 이상 row에 쓰이지 않으므로 제거한다.)
-
-- [ ] **Step 3: Modify backfill row to use build_run_index_row**
-
-[`scripts/okstra_ctl/backfill.py:219-235`](../../../scripts/okstra_ctl/backfill.py:219)의 `row = { ... }` 리터럴 전체를 치환:
-
-```python
-                    row = build_run_index_row(
-                        project_id=project_id, project_root=str(project_root),
-                        task_group=actual_group, task_id=actual_task,
-                        task_type=actual_type, run_seq=seq,
-                        status=normalized_status, started_at=started_at,
-                        finished_at=None if is_non_terminal else finished_at,
-                        workers=raw_workers, lead_model=raw_lead_model,
-                        validation=validation_status, run_dir_rel=run_dir_rel,
-                        final_report_rel=final_rel,
-                        final_status_rel=final_status_rel)
-```
-
-backfill.py 상단 import에 추가: `from .run_index_row import build_run_index_row`
-(라인 199-201의 `inv_rel` 계산은 row에서 더 이상 안 쓰이지만 save_invocation 에는 무관 — `inv_rel` 변수가 다른 곳에서 안 쓰이면 제거. 라인 244의 save_invocation은 `actual_group/actual_task/actual_type/seq`를 직접 넘기므로 영향 없음.)
-
-- [ ] **Step 4: Run regression tests**
-
-Run: `python3 -m pytest tests/run/ -v`
-Expected: PASS. 만약 기존 테스트가 디스크 row에서 `projectId`/`taskType`/`runSeq`/`taskKey`/`invocationFile`을 직접 단언한다면, 그 테스트의 read를 `read_run_index`로 바꾸거나 단언을 `hydrate_run_row(row)[...]` 기준으로 수정한다(assertion을 약화하지 말 것 — 동일 값 검증 유지).
-
-- [ ] **Step 5: Commit**
-
-```bash
-git add scripts/okstra_ctl/index.py scripts/okstra_ctl/backfill.py tests/run/
-git commit -m "refactor(index): writer 가 build_run_index_row 로 slim row 만 기록"
-```
-
----
-
-### Task 4: reader 배선 — sequence.py / listing.py
-
-**Files:**
-- Modify: `scripts/okstra_ctl/sequence.py:43-44`
-- Modify: `scripts/okstra_ctl/listing.py:84,88-89`
-- Test: 기존 `tests/run/` 회귀
-
-**Interfaces:**
-- Consumes: `read_run_index` (Task 2).
-- Produces: sequence/listing 의 row 소비 로직 무변경, 디스크 slim → 메모리 full.
-
-- [ ] **Step 1: sequence.py — read_jsonl → read_run_index**
-
-[`scripts/okstra_ctl/sequence.py:8-9`](../../../scripts/okstra_ctl/sequence.py:8) import 교체:
-`from .jsonl import read_jsonl` → `from .run_index_row import read_run_index`
-
-[`scripts/okstra_ctl/sequence.py:43-44`](../../../scripts/okstra_ctl/sequence.py:43):
-```python
-        for src in ("active.jsonl", "recent.jsonl"):
-            for r in read_run_index(home / src):
-```
-(아래 `r.get("projectId")`/`taskGroup`/`taskId`/`taskType`/`runSeq` 비교 로직은 그대로 — hydrate된 full row라 모두 존재.)
-
-- [ ] **Step 2: listing.py — list_runs 의 read 를 read_run_index 로**
-
-[`scripts/okstra_ctl/listing.py:10`](../../../scripts/okstra_ctl/listing.py:10) import 추가:
-`from .run_index_row import read_run_index` (기존 `from .jsonl import read_jsonl`는 다른 곳에서 안 쓰이면 제거).
-
-[`scripts/okstra_ctl/listing.py:84`](../../../scripts/okstra_ctl/listing.py:84):
-```python
-    rows = read_run_index(home / "active.jsonl") + read_run_index(home / "recent.jsonl")
-```
-
-[`scripts/okstra_ctl/listing.py:88-89`](../../../scripts/okstra_ctl/listing.py:88) (archive도 hydrate):
-```python
-            for f in sorted(archive_dir.glob("*/*.jsonl")):
-                rows.extend(read_run_index(f))
-```
-(`format_show`/`format_runs_table`/필터의 `projectId`/`taskType`/`invocationFile` 접근은 hydrate된 row라 그대로 동작.)
-
-- [ ] **Step 3: Run regression tests**
-
-Run: `python3 -m pytest tests/run/ -v`
-Expected: PASS. listing/sequence 테스트가 status/show 출력을 검증한다면 동일 출력 유지(파생 필드가 hydrate로 채워짐).
-
-- [ ] **Step 4: Commit**
-
-```bash
-git add scripts/okstra_ctl/sequence.py scripts/okstra_ctl/listing.py
-git commit -m "refactor(index): sequence/listing 이 read_run_index 로 hydrate 후 소비"
-```
-
----
-
-### Task 5: reconcile.py 배선 — read hydrate + write slim
-
-**Files:**
-- Modify: `scripts/okstra_ctl/reconcile.py` (read 4곳, write 4곳)
-- Test: 기존 `tests/run/test_*reconcile*.py` 회귀
-
-**Interfaces:**
-- Consumes: `read_run_index`, `slim_run_row` (Task 1·2).
-- Produces: reconcile 가 in-place 재작성 시 slim 유지. 종결 추론/promote 로직 무변경.
-
-- [ ] **Step 1: import 교체**
-
-[`scripts/okstra_ctl/reconcile.py:10`](../../../scripts/okstra_ctl/reconcile.py:10):
-```python
-from .jsonl import append_jsonl, rotate_recent_if_needed
-from .run_index_row import read_run_index, slim_run_row
-```
-(`read_jsonl` 제거 — 아래에서 모두 교체.)
-
-- [ ] **Step 2: read 지점 3곳 → read_run_index**
-
-- [`reconcile.py:89`](../../../scripts/okstra_ctl/reconcile.py:89) (`_sync_project_index`): `existing = read_jsonl(proj_index)` → `existing = read_run_index(proj_index)`
-- [`reconcile.py:121`](../../../scripts/okstra_ctl/reconcile.py:121): `rows = read_jsonl(active)` → `rows = read_run_index(active)`
-- [`reconcile.py:198`](../../../scripts/okstra_ctl/reconcile.py:198): `rows = read_jsonl(recent)` → `rows = read_run_index(recent)`
-
-(`row.get("projectId")`/`taskType`/`runSeq`/`runDirRel`/`projectRoot` 소비는 hydrate된 row라 그대로 동작.)
-
-- [ ] **Step 3: write 지점 4곳 → slim_run_row 적용**
-
-- [`reconcile.py:96-97`](../../../scripts/okstra_ctl/reconcile.py:96) (`_sync_project_index` 재작성):
-```python
-        for er in existing:
-            f.write(json.dumps(slim_run_row(er), separators=(",", ":")) + "\n")
-```
-- [`reconcile.py:170-171`](../../../scripts/okstra_ctl/reconcile.py:170) (active survivors 재작성):
-```python
-        for row in survivors:
-            f.write(json.dumps(slim_run_row(row), separators=(",", ":")) + "\n")
-```
-- [`reconcile.py:174`](../../../scripts/okstra_ctl/reconcile.py:174) (recent append):
-```python
-        append_jsonl(home / "recent.jsonl", slim_run_row(row))
-```
-- [`reconcile.py:223-225`](../../../scripts/okstra_ctl/reconcile.py:223) (recent in-place 재작성):
-```python
-        for row in rows:
-            f.write(json.dumps(slim_run_row(row), separators=(",", ":")) + "\n")
-```
-
-(라인 175·177의 `row["projectId"]`/`row["projectRoot"]`는 hydrate된 row라 존재 — 그대로.)
-
-- [ ] **Step 4: Run regression tests**
-
-Run: `python3 -m pytest tests/run/ -v`
-Expected: PASS (reconcile_active promote, reconcile_recent in-place, _sync_project_index 시나리오).
-
-- [ ] **Step 5: Commit**
-
-```bash
-git add scripts/okstra_ctl/reconcile.py
-git commit -m "refactor(index): reconcile read hydrate + write slim 로 정합 유지"
-```
-
----
-
-### Task 6: 일괄 마이그레이션 실행 + 빌드 + 전체 검증
-
-**Files:**
-- 변경 없음 (실행/검증 task)
-
-**Interfaces:**
-- Consumes: `reindex_slim` (Task 2).
-
-- [ ] **Step 1: 전체 단위 테스트**
-
-Run: `python3 -m pytest tests/`
-Expected: PASS (전 도메인).
-
-- [ ] **Step 2: 실 데이터 백업 후 마이그레이션 dry 확인**
-
-```bash
-cp -R ~/.okstra/recent.jsonl ~/.okstra/active.jsonl /tmp/okstra-index-backup/ 2>/dev/null; mkdir -p /tmp/okstra-index-backup
-cp ~/.okstra/recent.jsonl /tmp/okstra-index-backup/recent.jsonl
-python3 -c "import sys; sys.path.insert(0,'scripts'); from pathlib import Path; from okstra_ctl.run_index_row import reindex_slim, read_run_index; before=read_run_index(Path.home()/'.okstra'/'recent.jsonl'); print('rows', len(before))"
-```
-Expected: `rows 124` (또는 현재 수) — 에러 없이 hydrate 성공.
-
-- [ ] **Step 3: 마이그레이션 실행**
-
-```bash
-python3 -c "import sys; sys.path.insert(0,'scripts'); from pathlib import Path; from okstra_ctl.run_index_row import reindex_slim; print('migrated', reindex_slim(Path.home()/'.okstra'))"
-```
-Expected: `migrated <N>` (recent + active + 6 index 의 총 row 수, 약 246).
-
-- [ ] **Step 4: 마이그레이션 후 정합 검증**
-
-```bash
-python3 -c "import sys,json; sys.path.insert(0,'scripts'); from pathlib import Path; from okstra_ctl.run_index_row import read_run_index; bak=[json.loads(l) for l in open('/tmp/okstra-index-backup/recent.jsonl') if l.strip()]; cur=read_run_index(Path.home()/'.okstra'/'recent.jsonl'); bak_by={r['runId']:r for r in bak}; assert all(cur_r['invocationFile']==bak_by[cur_r['runId']].get('invocationFile') and cur_r['taskKey']==bak_by[cur_r['runId']].get('taskKey') for cur_r in cur if cur_r['runId'] in bak_by), 'mismatch'; print('OK: hydrate 가 원본 파생필드 재현')"
-```
-Expected: `OK: hydrate 가 원본 파생필드 재현` — 마이그레이션 전 full row의 파생 필드와 hydrate 결과가 일치.
-
-- [ ] **Step 5: 멱등 재실행 + CLI 스모크**
-
-```bash
-python3 -c "import sys; sys.path.insert(0,'scripts'); from pathlib import Path; from okstra_ctl.run_index_row import reindex_slim; print('reidempotent', reindex_slim(Path.home()/'.okstra'))"
-node bin/okstra --version
-```
-Expected: 두 번째 reindex_slim도 정상, CLI 정상.
-
-- [ ] **Step 6: runtime 빌드 + 커밋**
-
-```bash
-npm run build
-git add -A
-git commit -m "chore(index): runtime 동기화 (run-index row slim 포맷)"
-```
-Expected: `runtime/` 가 새 소스로 갱신.
-
----
-
-## Self-Review
-
-- **Spec coverage:** §3.1 제거 5필드 → Task 1 `DERIVED_FIELDS`/`build_run_index_row`. §3.2 유지 13필드 → Task 1 row 구성 + 테스트. §4.1 두 헬퍼 → Task 1·2. §4.2 read hydrate(6곳 후보 → 검증으로 3곳 확정) → Task 4·5 (pane_reclaim/recap/wizard는 run-index row 비소비로 제외, 본문에 근거 명시). §5 마이그레이션(migrate.py 분리) → Task 2 `reindex_slim` + Task 6 실행. §6 테스트 5종 → Task 1·2 테스트. §7 비범위 → 계획에 포함 안 함.
-- **Placeholder scan:** 모든 코드 step에 실제 코드/명령/기대출력 포함. TBD 없음.
-- **Type consistency:** `build_run_index_row`/`slim_run_row`/`hydrate_run_row`/`read_run_index`/`reindex_slim`/`DERIVED_FIELDS` 시그니처가 Task 1→6 전반에서 일관. row 키 camelCase 일관.
-- **주의(실행 중 확인):** 기존 `tests/run/`의 일부 테스트가 디스크 row에서 파생 필드를 직접 단언할 수 있다. Task 3·4·5의 회귀 step에서 발견 시, read를 `read_run_index`로 바꾸거나 단언을 동일 값 기준으로 갱신한다 — assertion 약화 금지.