本文へ移動
cccskills
無料GitHub で公開

run-dev-graph-status

dev-graph node を metadata で検索したいとき、依存・tombstone・completion 状態を副作用なしで確認したいときに使う。

インストール方法を見る

含まれるファイル(4)

  • SKILL.md9.6 KB
  • prompts/R1-elicit.md2.5 KB
  • prompts/R2-plan.md2.4 KB
  • prompts/R3-status.md2.8 KB

SKILL.md(原文)

インストールする前に、エージェントに与えられる指示の中身を確認できます。

Pre-choice usable artifact execution

Purpose & Output Contractの最小の実成果物をmain contextで作成する。effect別のparse/open・secret・irreversible・corrupt guardだけを実行し、現物path・digest・開き方を提示してからaccept-as-is/light/standard/detailedを記録する。accept-as-isはその場でhandoff完了とし、後続sectionを実行しない。

Post-choice selected improvement execution

以下の既存workflow・goal-seek・評価・修正sectionはlight/standard/detailedが記録されてsemantic_evaluator_startedへ遷移した場合だけ実行する。release/exhaustiveは別の明示eventを必要とする。

run-dev-graph-status

Purpose & Output Contract

  • 入力: C24/C11 で read-only 検証済み graph store、scope/filter/tag match mode。
  • 出力: node identity、classification、path、status/closed_at、依存関係、feature/tracker linkage を持つ検索 report。
  • 完了条件: report が graph 実状態と一致し、実行前後の graph/config/content digest が不変で外部副作用0である。

C24 で caller root を固定し、C11 の read-only validation 後に graph store を検索する。全 filter は AND、複数 tag は明示された match mode に従う。結果は graph_node_id/artifact_kind/project_id/domain/tags/file_path/status/closed_at/depends_on/dependents/parent_feature/feature_package_id/tracker_binding/linkage を返す。

検索前後の graph/config/content digest を比較し、変化があれば失敗。GitHub/Beads command、writer、render、sync を呼ばない。該当なしは空 result の成功、schema 不正・root 外 file_path・dangling dependency は診断付き失敗。

ゴールシーク実行

ゴール (Goal)

特定のissue/task/specification/architecture/documentをgraph_node_id/artifact_kind/project_id/domain/status/tagsで検索し、依存関係・close/tombstone状態をread-onlyで確認できる状態になっている

目的・背景 (Why)

特定ノードの検索・状態確認 (要件C11) を独立の読取り専用経路として提供しないと、状態確認のたびに sync/render 相当の重い処理を経由せざるを得ず、副作用のない検索操作が書込みを伴う操作と混同されるリスクがあるため

完了チェックリスト

  • query は許可 filter field と明示 tag match mode だけを持つ
  • validate-graph-schema.py が exit0 で root 外 file_path と dangling dependency が0件である
  • report の status/closed_at/depends_on/dependents/linkage が graph snapshot と一致する
  • 該当なしが空 result の成功として返る
  • 実行前後の graph/config/content digest が同一で外部 write が0件である

ゴールシークループ

frontmatter の goal_seek.engine: inline / fork: subagent / max_loops: 5 を実行契約とする。固定手順は使わず、未達 checklist と担当 prompts/*.md からその周回の操作を都度生成する。各周回で inner criterion を検証し、完了後は outer criterion の live trial/content review を最大 feedback_contract.max_iterations=3 周で評価する。

ゴールシーク配線

  • 開始時に C24 resolve-repo-context.py --mode read の JSON receipt を得て、repo_root が content_roots.repository の realpath と一致する場合だけ DEV_GRAPH_ROOT=<receipt.repo_root> に固定する。cwd から再解決しない。
  • 元のゴールを $DEV_GRAPH_ROOT/eval-log/run-dev-graph-status-goal-spec.json へ、各 checklist の status/evidence を $DEV_GRAPH_ROOT/eval-log/run-dev-graph-status-progress.json へ記録する。
  • 未達 responsibility を担当する prompts/<R-id>.md を読み、Agent で分離 context に fork する。ユーザー判断が必要な境界だけ AskUserQuestion を使う。
  • 各周回末に $DEV_GRAPH_ROOT/eval-log/run-dev-graph-status-intermediate.jsonl へ original_goal、original_goal_hash、current_goal_snapshot、delta_from_original、merged_directive_for_next、drift_signal を append-only で記録する。次周回は直前の merged_directive_for_next を必須入力にする。
  • 5周到達時に未達が残れば完了扱いせず、progress と blocker を親へ handoff する。全 checklist と feedback_contract.criteria が PASS のときだけ完了する。

ゴールシーク検証

各周回後に次の検査を実行し、中間成果物の欠落・goal drift・hash 不一致を fail-closed にする。

python3 - "$DEV_GRAPH_ROOT/eval-log/run-dev-graph-status-goal-spec.json" "$DEV_GRAPH_ROOT/eval-log/run-dev-graph-status-intermediate.jsonl" <<'PY'
import hashlib, json, sys
goal = json.load(open(sys.argv[1], encoding='utf-8'))
rows = [json.loads(line) for line in open(sys.argv[2], encoding='utf-8') if line.strip()]
required_keys = {'original_goal','original_goal_hash','current_goal_snapshot','delta_from_original','merged_directive_for_next','drift_signal'}
expected = hashlib.sha256(goal['original_goal'].encode('utf-8')).hexdigest()
assert rows, 'intermediate.jsonl is empty'
for row in rows:
    assert required_keys <= row.keys(), required_keys - row.keys()
    assert row['original_goal'] == goal['original_goal']
    assert row['original_goal_hash'] == expected
PY

Criteria acceptance

  • criteria:IN1: validate-graph-schema.py後の結果は必須キー欠落が0件である。
  • criteria:OUT1: reportのstatus/closed_at/依存関係はgraphの実状態と一致する。
  • criteria:OUT2: 実行前後のdigestを比較し、graph/content/GitHubの副作用0件を保証する。

Gotchas

  • filter 間は OR に緩めず、全て AND で適用する。複数 tag は指定された match mode に従う。
  • node 本文から status を推測せず、graph の status/closed_at をそのまま報告する。
  • 該当なしは空 result の成功とし、schema 不正や dangling dependency と混同しない。
  • writer、sync、render、GitHub/Beads command を呼び出さない。

レビュー

まだレビューはありません。使ってみた感想をお寄せください。

同じリポジトリのスキル

概要と使いどころ

app-excellence

無料日本語概要

アプリ・Webサービス・業務ツールの制作を、証拠分析→確定的な初稿→要件定義→体験設計→機能分解→実装→品質ゲート→リリース判定の手順ゲート制で進めるスキル。質問で要件を埋めるのではなく、依頼文・既存コード・資料・ログから最有力案を選び、動く成果物を先に出して差分で改善する。ユーザーが「アプリを作りたい」「ツールを作って」「〜を自動化したい」「システム化したい」「要件定義」「仕様を決めたい」「UI/UXを設計して」「体験を良くしたい」「機能を洗い出して」「MVPを決めたい」「品質チェック」「リリースしていいか判断して」などと言ったら必ず使用する。外部データの取込・マスタ・締め処理を伴う業務システムでは references/data-lifecycle.md も併用する。新規・既存を問わずアプリの制作・改善・リニューアルで使用する。

daishiman/harness-dev102026年10月10日 更新

app-orchestrator

無料日本語概要

Webアプリの準備→要件定義→設計→実装→公開→品質ゲートを実行する内部オーケストレーター。Claude Codeの /build-app /improve-app、Codexの $build-app / $improve-app から明示的に委譲された場合、custom agent起動時、または利用者が $app-orchestrator を明示した場合だけ使用する。一般のアプリ相談から暗黙起動しない。

daishiman/harness-dev102026年10月10日 更新

run-extract-blueprint が生成した章別ブループリントの忠実性を独立 context で評価したいとき、事実/推測区別と粒度と被覆を検証し draft_hash に束縛した PASS/FAIL verdict をローカル品質ゲート (C01 の周回内の受入判定・差し戻し) へ渡したいときに使う。

daishiman/harness-dev102026年10月10日 更新

assign-briefing-evaluator

無料日本語概要

確認点で利用者が見る深さを選んだあと打ち合わせ資料を作り手とは別の目で確かめたいとき、正確さ・言葉・見た目・シンプルさの指摘を資料の編集なしで受け取りたいときに使う。

daishiman/harness-dev102026年10月10日 更新

生成した handout 資料が初心者に伝わるか読みやすさレビューを依頼したいとき、独立 context のレビュアーから指摘と根拠つきの verdict を回収したいときに使う。

daishiman/harness-dev102026年10月10日 更新

Notion ページを描画する直前に粒度を検証したいとき、info-collector-agent ページと同等の section 充足度を section_canonical_map 基準で機械検証したいときに使う。

daishiman/harness-dev102026年10月10日 更新

daishiman のスキルをすべて見る

このスキルの問題を報告する