deep-status

Files (4)

• SKILL.md
• compare-mode.md · references/
• flag-views.md · references/
• fork-views.md · references/

SKILL.md (first lines)

## Inputs (skill args)

| 인자 | 의미 |
|---|---|
| (없음) | Default status dashboard |
| `--compare` | Fork session 비교 (parent vs forks) |
| `--receipts` | deep-receipt sub-skill 호출 (§6) |
| `--history` | deep-history sub-skill 호출 (§7) |
| `--report` | deep-report sub-skill 호출 (§8) |
| `--assumptions` | deep-assumptions sub-skill 호출 (§9) |
| `--risk` | Shadow risk/policy 표시 (§13) |
| `--all` | 4 sub-page 모두 순차 실행 |
| `--tree` | Fork tree 출력 |
| `--badge` | Shields.io 호환 badge 출력 |

빈 args / 매칭되지 않는 토큰 → 본문의 default 분기로 진입.

## Prerequisites

**Hub-spoke 관계**: 본 skill 은 `deep-receipt` / `deep-history` / `deep-report` / `deep-assumptions` 4 개 sub-skill 의 hub 입니다 — 본문 §6 ~ §9 가 각 sub-skill 의 SKILL.md 를 inline Read 하여 로직을 실행합니다. 4 sub-skill 은 standalone 으로도 호출 가능합니다.

# Deep Work Status

Display the current state of the Deep Work session and session history.

## Language

Read(`${CLAUDE_PLUGIN_ROOT}/skills/shared/references/user-language.md`) and follow it.

## Instructions

### 0. Check for compare mode

`$ARGUMENTS`에 `--compare`가 있으면
`${CLAUDE_PLUGIN_ROOT}/skills/deep-status/references/compare-mode.md`
를 읽고 그대로 수행한 뒤 종료한다. 없으면 §0-1로 진행한다.

### 0-1. Parse flags

Parse `$ARGUMENTS` for the following flags. If multiple flags are provided, execute each in order.

| Flag | Effect |
|------|--------|
| `--receipts` | Show receipt dashboard |
| `--receipts SLICE-NNN` | Show specific slice receipt detail |
| `--receipts --export=json` | Export all receipts as single JSON |
| `--receipts --export=md` | Export as markdown (for PR descriptions) |
| `--receipts --export=ci` | Export CI bundle |
| `--history` | Show cross-session trends |
| `--report` | Show/generate session report |
| `--assumptions` | Show assumption health report |
| `--assumptions --verbose` | Per-signal per-session breakdown |
| `--assumptions --rebuild` | Regenerate JSONL from receipts, then show report |
| `--risk` | Show shadow risk profile & policy recommendation |
| `--badge` | Generate shields.io badge markdown |
| `--tree` | Fork relationship tree visualization |
| `--all` | Show all sessions dashboard (multi-session) + all flags |
| `--compare` | Compare two sessions (existing, handled in Section 0) |

If no flags are provided (and no `--compare`), show the default view only (Steps 1-5).
If a flag is provided, execute the corresponding section after the default view.

### 1. Check if a session exists (multi-session aware)

Read(`${CLAUDE_PLUGIN_ROOT}/skills/deep-resume/references/session-detection.md`) and apply only its **Reusable session-state resolution** section to resolve `$STATE_FILE`; retain this skill's own no-session and standalone-mode behavior.

If resolution returns `none` or `invalid-explicit`, display:

```
ℹ️ 활성화된 Deep Work 세션이 없습니다.

새 세션을 시작하려면: /deep-work <작업 설명>
```

If flags were provided (`--history`, `--assumptions`, `--receipts`, `--report`, `--all`):
- Skip the default view (Steps 2-4) but still execute the corresponding flag handler sections (Steps 6-10). These features can work without an active session by reading historical data from `.deep-work/` directory.

If no flags were provided:
- Skip to [Step 5: Show session history](#5-show-session-history).

### 2. Read state and artifacts

Read the resolved state file (from Step 1) to get session state.

Extract `work_dir` from the state file. If missing, default to `deep-work` (backward compatibility).
Set `WORK_DIR` to this value.

Read(`${CLAUDE_PLUGIN_ROOT}/skills/shared/references/model-routing-guide.md#model-routing-state-decode-v612`)의
scalar-first 규칙으로 `decodedRouting`을 만든다. decode 실패 시 기본값
(Research=sonnet, Plan=현재 세션, Implement=sonnet, Test=haiku)을 표시한다.

Read `evaluator_model`, `assumption_adjustments`, `skipped_phases`, `plan_review_retries`, and `plan_review_max_retries` from the state file. If missing, default to: evaluator_model="없음", assumption_adjustments=[] (empty), skipped_phases=[] (empty), plan_review_retries=0, plan_review_max_retries=3.

Read the following files if they exist:
- `$WORK_DIR/research.md` — check if it has content
- `$WORK_DIR/plan.md` — count checklist progress
- `$WORK_DIR/report.md` — check if it exists
- `$WORK_DIR/test-results.md` — check if it exists
- `$WORK_DIR/quality-gates.md` — check if it exists
- `$WORK_DIR/insight-report.md` — check if it exists
- `$WORK_DIR/file-changes.log` — check if it exists
- `$WORK_DIR/plan-diff.md` — check if it exists
- Read `review_state`, `cross_model_enabled`, and `review_results` from state file
- Review finding은 `readFindings({workDir, point, round})`로 읽는다. canonical reviews 경로를
  우선하며 함수가 제공하는 legacy fallback(`phase-cross-review.json` →
  `adversarial-review.json`)과 `source:'legacy'` 표기를 그대로 표시한다. structural 호환
  `${phase}-review.json`은 점수 요약에만 사용한다.

### 2-1. Read sensor data

If receipts directory exists (`$WORK_DIR/receipts/`):
- Read all `SLICE-NNN.json` receipt files
- **Envelope-aware unwrap**: 각 receipt 의 root 가 M3 envelope
  형태(`schema_version === "1.0"` + `envelope` 객체 + `payload` 키)이면
  identity guard 검증 (`envelope.producer === "deep-work"` ∧
  `envelope.artifact_kind === "slice-receipt"` ∧
  `envelope.schema.name === "slice-receipt"`) 후 `payload` 를 receipt body
  로 사용. mismatch / corrupt payload 는 "foreign envelope" 경고 후 무시.
  Legacy(non-envelope) receipt 는 root 객체를 그대로 사용 (forward-compat).

From Sungmin-Cho/deep-work, MIT.

More from Sungmin-Cho/deep-work