feat(notion): add v2 work report platform

.gitignore

@@ -28,3 +28,6 @@ claude-code-working-diary.tar.gz
 
 # /diary-notion 임시 JSON (Claude가 cwd에 작성 → CLI가 자동 삭제 / 부분 실패 시 보존)
 .diary-notion-*.json
+
+# Local v2 diary-notion run artifacts
+.codefleet/runs/

CHANGELOG.md

@@ -24,6 +24,8 @@ and this project adheres to [Semantic Versioning](https://semver.org/).
 - **Working Diary OS 비전 문서**: Structure → Views → Operations → Intelligence → Multi-project OS로 확장하는 최고모델 설계, 최소 명령 원칙, 전날 todo 기반 `today-plan`, schema/view conflict drift 관리 방향 추가
 - **2차 View 설계 문서**: `working-diary diary-notion ensure`, `--year`, `--dry-run`과 Core Views 5개, operating views 5개, `Work Period`와 `Sub-items` 기반 schema v7 방향, 하위 항목 데이터 구조, sub-item UI best-effort/fallback, partial failure/exit code 정책 정리
 - **`working-diary diary-notion ensure` 구현**: schema v7 `Work Period`, native sub-item relation, Priority/Blocked/Review 운영 컬럼 보장, Core Views 5개와 Operating Views 5개 생성/검증/update, `--year`, `--dry-run`, required setting repair 지원
+- **`working-diary diary-notion ops` 구현**: Notion Entries DB를 읽기 전용으로 조회해 blocked/review/next action/stale/work days/today-plan 후보/task group/project/parent progress/부모 상태 제안 운영 신호를 요약
+- **Phase 3 ensure diagnostics**: schema/view conflict reason을 `missing_filter`, `missing_property`, `subitem_missing`, `permission_or_auth`, `api_failure`로 분류하고 repair plan/action/apply 가능 여부를 출력
 - **`claude-diary write --input <file>`**: Codex skill이 생성한 JSON으로 수동 Markdown 일지 작성
 - **`lib/notion_cache.py`**: 연도 페이지/DB/행 ID 캐시 (root_page_id 변경 시 자동 무효화)
 - **`lib/git_info.py` 확장**: `get_branch_for_commit`, `get_head_branch`, `get_commit_info`, `get_diff_stat_for_commits`

README.en.md

@@ -335,6 +335,7 @@ working-diary write
 working-diary diary-notion init
 working-diary diary-notion ensure
 working-diary diary-notion ensure --dry-run
+working-diary diary-notion ops
 working-diary diary-notion push --input .diary-notion-<id>.json
 working-diary notion push --input .diary-notion-<id>.json
 ```
@@ -407,6 +408,7 @@ $OutputEncoding = [System.Text.UTF8Encoding]::new()
 - Secret scanning and masking
 - Search index
 - Notion work log: `working-diary diary-notion init` -> `working-diary diary-notion ensure`
+- Notion operations report: `working-diary diary-notion ops` for blocked/review/next action/stale/work days/today-plan candidates/parent status signals
 - Slack, Discord, Obsidian, GitHub exporters: `working-diary config --add-exporter <name>`
 - HTML dashboard: `working-diary dashboard` or `working-diary dashboard --serve --port 8787`
 - Audit log and source checksum verification
@@ -438,6 +440,7 @@ This README focuses on currently usable functionality. Detailed design and plann
 | Area | Status |
 |------|--------|
 | Stable now | Claude Code Stop Hook, Codex skills, Markdown diaries, Notion task row push, schema/view ensure |
+| Phase 3 started | `working-diary diary-notion ops` read-only operations report, parent/task group progress, ensure conflict classification and repair plans |
 | Next | Windows install/output experience, Notion sub-item guidance, incremental CI/lint expansion |
 | Under review | SQLite search index, Cursor/Windsurf/VS Code integration |
 

README.md

@@ -339,6 +339,7 @@ working-diary write
 working-diary diary-notion init
 working-diary diary-notion ensure
 working-diary diary-notion ensure --dry-run
+working-diary diary-notion ops
 working-diary diary-notion push --input .diary-notion-<id>.json
 working-diary notion push --input .diary-notion-<id>.json
 ```
@@ -411,6 +412,7 @@ $OutputEncoding = [System.Text.UTF8Encoding]::new()
 - secret scan과 masking
 - 검색 인덱스
 - Notion 업무일지: `working-diary diary-notion init` -> `working-diary diary-notion ensure`
+- Notion 운영 진단: `working-diary diary-notion ops`로 blocked/review/next action/stale/work days/today-plan 후보/부모 상태 제안 확인
 - Slack, Discord, Obsidian, GitHub exporter: `working-diary config --add-exporter <name>`
 - HTML dashboard: `working-diary dashboard` 또는 `working-diary dashboard --serve --port 8787`
 - audit log와 source checksum 검증
@@ -442,6 +444,7 @@ python -m ruff check .
 | 구분 | 내용 |
 |------|------|
 | 현재 안정화 | Claude Code Stop Hook, Codex skill, Markdown 일지, Notion task row push, schema/view ensure |
+| Phase 3 진입 | `working-diary diary-notion ops` 읽기 전용 운영 진단, parent/task group 진행률, ensure conflict 분류와 repair plan |
 | 다음 개선 | Windows 설치/출력 경험 정리, Notion sub-item 안내 개선, CI/lint 범위 점진 확대 |
 | 검토 중 | SQLite 기반 검색 인덱스, Cursor/Windsurf/VS Code 같은 다른 AI IDE 연동 |
 

skills/diary-notion/SKILL.md

@@ -10,38 +10,47 @@ Split the current Codex session into task-sized entries and push them to Notion.
 ## Current Implementation Contract
 
 - `$diary-notion` is a row push workflow only. Do not run schema/view ensure unless the user explicitly asks for it.
-- `working-diary diary-notion ensure` is the separate maintenance command that guarantees schema v7, native sub-items, 5 core views, and 5 operating views.
+- `working-diary diary-notion ensure` is the separate maintenance command that guarantees schema v8, native sub-items, 5 core views, and 5 operating views.
 - Use Notion native sub-items for containment by setting `parent_index`. If the native relation is not enabled in Notion, still push the rows and report that sub-item activation is needed.
 - Treat legacy `Parent Task` / `Sub-items` as compatibility data only. Do not target them directly in JSON.
 - Use `Depends On` only for prerequisite links between large top-level main tasks. Never use dependencies for child tasks.
 - Never write `"unknown"` as `project`; omit it or leave it blank so the CLI falls back to the command cwd folder name.
-- Page bodies render as compact executive bodies: top summary, result checklist, work-at-a-glance table, impact, verification, risks/next action, and appendix.
+- Page bodies render as compact work reports: summary, results, work table, decisions, issues/risks, next actions/support, and appendix toggles.
+- Notion is the report surface; raw logs, long diffs, and bulky evidence belong in local artifact files and should be referenced by path/hash instead of pasted.
+- `diary-notion push` writes local run artifacts under `.codefleet/runs` by default: `input.json`, `git-diff.patch`, `preview.md`, and `manifest.json`. Use `--preview-file` for an extra Markdown preview path or `--no-artifacts` to disable local artifact writes.
+- For testing, QA, review, validation, or verification sessions, create a row even without code changes; keep `verification` short and place the meaningful prompt-result document in `prompt_outputs` or `verification_artifacts` so it renders inside a toggle.
 
 ## Workflow
 
 1. Review the current conversation, tool calls, git branch, and relevant git commits.
 2. Split work into task-sized database rows. Branch changes are hard task boundaries; within a branch, split by semantic work unit.
    - Create a row for work that has its own status, evidence, code/test output, or can block another task
+   - Create a row for verification-only work even when the session produced no file changes or commits
    - Keep tiny check items, raw notes, long SQL/JS snippets, and reference links inside the page body evidence instead of making them separate rows
    - Create a separate row only when the work has an independent status, verification/evidence, code change, commit, blocker, or follow-up owner
+   - For tester/verification sessions, summarize the final state in `verification` and put each meaningful pass, fail, blocker, skipped check, regression, defect, and follow-up in the prompt-output artifact fields
    - Use `parent_index` for containment hierarchy and Notion sub-items; do not model subtasks with dependencies
    - Use `depends_on_indices` only for prerequisite links between large top-level tasks
    - Mark continued work from an earlier day/session as a new row with the same `task_group` and `carryover=true` when it is still unfinished
 3. For each task, produce:
+   - Prefer schema v2 normalized fields: `summary`, `work`, `decisions`, `risks`, `next_actions`, `support_needed`, and `appendix`.
+   - Legacy flat fields are still accepted by the CLI, but new agent output should use v2 unless compatibility with an older installed version is required.
    - Language policy:
-     - Write `title`, `body_intro`, `summary_hints`, `key_changes`, `work_context`, `work_scope`, `approach`, `outcome`, `impact`, `decisions`, `implementation_notes`, `verification`, `risks`, `next_steps`, `support_needed`, `next_action`, and `block_reason` in Korean
+     - Write `title`, `summary`, `work`, `decisions`, `risks`, `next_actions`, `support_needed`, `appendix`, `next_action`, and `block_reason` narrative values in Korean
      - Keep `status` and `purpose` as the exact English enum values below
      - Preserve file paths, commands, branches, commit hashes, code identifiers, function names, and class names as written
      - Preserve `user_prompts` in the user's original wording as evidence
    - `title`: concise Korean noun phrase, no prefix or period
    - `body_intro`: 1-3 factual Korean sentences based only on observed work
    - Write it like a Notion task database record: compact top summary, structured relations in DB properties, and raw evidence hidden in the page body appendix
    - Body rendering policy:
-     - Use `body_intro` as the only top summary callout
-     - Treat `summary_hints` as checked result items, not repeated callouts
-     - Keep `work_context`, `work_scope`, `approach`, and `outcome` short because they render as a compact "work at a glance" table
-     - Put final verification state in `verification`; move intermediate command history to appendix evidence
-     - Keep risks concise; multiple risks are combined into one warning callout
+     - Use `summary.intro` as the only top summary callout
+     - Treat `summary.outcomes`, `summary.verification`, and `summary.remaining` as the compact results checklist
+     - Keep `work.context`, `work.scope`, `work.approach`, and `work.state` short because they render as a compact work table
+     - Put settled choices in `decisions`; put unresolved external asks in `support_needed`
+     - Keep `risks` concise as issue/risk bullets, not raw logs
+     - Put developer evidence, prompt outputs, original requests, and command/file/commit evidence in `appendix` toggles
+     - For tester/verification sessions, keep `summary.verification` to 1-3 summary items and put the meaningful prompt-result document in `appendix.prompt_outputs` or `appendix.verification_artifacts`; summarize long raw logs instead of pasting them
    - `summary_hints`: up to 3 outcome-focused result items that explain what changed and why it matters
    - `key_changes`: up to 3 major behavior/schema/workflow changes a developer can understand without opening the diff
    - `work_context`: 0-1 bullet explaining why this work started
@@ -55,12 +64,16 @@ Split the current Codex session into task-sized entries and push them to Notion.
      - Include changes to behavior, schema, CLI flow, user workflow, or verification scope
    - `decisions`: 0-3 decisions or tradeoffs made by the user or settled during implementation
    - `implementation_notes`: 0-4 constraints, compatibility notes, migrations, or details that do not fit code highlights
-   - `verification`: 0-3 tests/checks run, results, or explicit reasons checks were not run
+   - `verification`: 0-3 summary tests/checks run, final results, or explicit reasons checks were not run
+   - `prompt_outputs`: 0-15 meaningful prompt-result items for tester/verification sessions; include distinct pass/fail/blocker/skipped/regression/defect/follow-up results, not raw logs
+   - `verification_artifacts`: 0-15 structured verification artifact items when the prompt output is better grouped as a generated test report or review document
+   - `appendix.artifacts`: 0-5 local artifact references with `path`, `kind`, `summary`, and optional `sha256`; use this for stdout/stderr/diff/raw-log evidence
    - `risks`: 0-2 cautions, remaining risks, or usage/operation notes
    - `next_steps`: 0-2 remaining follow-ups
    - `support_needed`: 0-1 decisions or support needed from others
    - `status`: `Discussion`, `Design`, `Implementation`, `Testing`, or `Deployed`
    - `purpose`: `Feature`, `Bugfix`, `Refactor`, `Docs`, `Test`, `Infra`, `Planning`, `Research`, `Review`, `Release`, `Support`, `Maintenance`, or `General`
+     - Use `Test` for tester, QA, validation, and verification-only sessions unless another enum is clearly more accurate
    - `work_period`: actual work period; use today's `YYYY-MM-DD` by default, or `{"start":"YYYY-MM-DD","end":"YYYY-MM-DD"}` for a range
    - `priority`: one of `P0`, `P1`, `P2`, `P3`; use `P0` for urgent/blocking work, `P1` for today's highest priority, `P2` for normal follow-up, and `P3` for low priority
    - `next_action`: 0-1 concrete Korean action that can be started next
@@ -81,24 +94,46 @@ Split the current Codex session into task-sized entries and push them to Notion.
 ```json
 {
   "session_id": "<current session id or codex-manual>",
+  "schema_version": 2,
   "tasks": [
     {
       "title": "...",
-      "body_intro": "...",
-      "summary_hints": ["..."],
-      "key_changes": ["..."],
-      "work_context": ["..."],
-      "work_scope": ["..."],
-      "approach": ["..."],
-      "outcome": ["..."],
-      "impact": ["..."],
-      "code_change_highlights": ["..."],
+      "summary": {
+        "intro": "...",
+        "outcomes": ["..."],
+        "verification": ["..."],
+        "remaining": ["..."]
+      },
+      "work": {
+        "context": "...",
+        "scope": "...",
+        "approach": "...",
+        "state": "..."
+      },
       "decisions": ["..."],
-      "implementation_notes": ["..."],
-      "verification": ["..."],
       "risks": ["..."],
-      "next_steps": ["..."],
+      "next_actions": ["..."],
       "support_needed": ["..."],
+      "appendix": {
+        "key_changes": ["..."],
+        "implementation_notes": ["..."],
+        "prompt_outputs": ["..."],
+        "verification_artifacts": ["..."],
+        "user_prompts": ["..."],
+        "files_modified": ["..."],
+        "files_created": ["..."],
+        "commands_run": ["..."],
+        "commit_hashes": ["..."],
+        "errors": ["..."],
+        "artifacts": [
+          {
+            "kind": "stdout",
+            "path": ".codefleet/runs/<run-id>/stdout.log",
+            "summary": "raw command output",
+            "sha256": "..."
+          }
+        ]
+      },
       "status": "Implementation",
       "purpose": "Feature",
       "work_period": "2026-06-02",
@@ -113,20 +148,16 @@ Split the current Codex session into task-sized entries and push them to Notion.
       "parent_index": null,
       "depends_on_indices": [],
       "categories": ["feature"],
-      "project": "<cwd folder name>",
-      "user_prompts": ["..."],
-      "files_modified": ["..."],
-      "files_created": ["..."],
-      "commands_run": ["..."],
-      "commit_hashes": ["..."],
-      "errors": ["..."]
+      "project": "<cwd folder name>"
     }
   ]
 }
 ```
 
-5. Run `working-diary diary-notion push --input .diary-notion-<8-random>.json`.
-6. If `working-diary` is not available, run `claude-diary diary-notion push --input .diary-notion-<8-random>.json`.
-7. Report pushed/skipped/failed tasks from the CLI output.
+5. Run `working-diary diary-notion push --input .diary-notion-<8-random>.json --dry-run` to validate v2 input, write local artifacts, and preview the compact report body and appendix toggles without writing to Notion.
+6. If the preview is structurally wrong, fix the JSON before pushing.
+7. Run `working-diary diary-notion push --input .diary-notion-<8-random>.json`.
+8. If `working-diary` is not available, run `claude-diary diary-notion push --input .diary-notion-<8-random>.json`.
+9. Report pushed/skipped/failed tasks from the CLI output.
 
 If there are no task-worthy changes, explain that and do not call the CLI.

src/claude_diary/cli/__init__.py

@@ -23,6 +23,7 @@
 from claude_diary.cli.notion_push import cmd_notion_push
 from claude_diary.cli.notion_init import cmd_notion_init
 from claude_diary.cli.notion_ensure import cmd_notion_ensure
+from claude_diary.cli.notion_ops import cmd_notion_ops
 
 
 def cmd_notion(args):
@@ -33,17 +34,29 @@ def cmd_notion(args):
         cmd_notion_init(args)
     elif args.action == "ensure":
         cmd_notion_ensure(args)
+    elif args.action == "ops":
+        cmd_notion_ops(args)
 
 
 def _add_diary_notion_parser(sub, name):
     p_notion = sub.add_parser(name, help="Notion hierarchical work diary integration")
-    p_notion.add_argument("action", choices=["init", "push", "ensure"], help="Action to perform")
+    p_notion.add_argument("action", choices=["init", "push", "ensure", "ops"], help="Action to perform")
     p_notion.add_argument("--input", help="JSON input file (push only)")
     p_notion.add_argument("--force", action="store_true",
                           help="Archive prior rows for the session before pushing (push only)")
-    p_notion.add_argument("--year", type=int, help="Target year (ensure only)")
+    p_notion.add_argument("--year", type=int, help="Target year (ensure/ops only)")
     p_notion.add_argument("--dry-run", action="store_true",
-                          help="Print the Notion schema/view plan without writing (ensure only)")
+                          help="Preview push body without writing, or print ensure plan")
+    p_notion.add_argument("--preview-file",
+                          help="Write push dry-run/preview Markdown to this file")
+    p_notion.add_argument("--artifact-dir", default=".codefleet/runs",
+                          help="Directory for local run artifacts (push only)")
+    p_notion.add_argument("--no-artifacts", action="store_true",
+                          help="Do not write local run artifacts (push only)")
+    p_notion.add_argument("--stale-days", type=int, default=7,
+                          help="Mark active rows stale after N days (ops only)")
+    p_notion.add_argument("--json", dest="json_output", action="store_true",
+                          help="Output operations report as JSON (ops only)")
 
 
 def main():