feat(cli-push-runner): Phase c MVP — pre-push lint-screen step (default OFF) by aloekun · Pull Request #132 · aloekun/claude-code-hook-test

aloekun · 2026-05-08T19:01:15Z

Summary

Phase c MVP — cli-push-runner に lint-screen step を追加し、pre-push 時に diff を mistral:7b に流して lint 一次フィルタの所見 (.takt/lint-screen-report.md) を出力する経路を整備。Phase b' conditional GO (75%) のため gating せず、report 出力のみ (default OFF / 試験運用 opt-in)。

アーキテクチャ

quality_gate → diff (.takt/review-diff.txt) → [lint_screen]? → takt → push

配置: cli-push-runner (Rust) の新 stage stages/lint_screen.rs
起動条件: [lint_screen] enabled = true 時のみ (default OFF)
subprocess: cli-finding-classifier.exe --mode lint-screen (stdin に diff、stdout に LintScreenResult JSON)
出力: .takt/lint-screen-report.md (markdown table)
連携: review-simplicity.md instruction が同 path を optional context として読む (advisory 扱い)

fallback 設計 (push を block しない)

exe 不在 → skip + warn
diff 空 → skip
diff 過大 (max_diff_lines = 5000、default) → skip + warn
timeout (timeout_secs = 60、default) → skip + warn
Ollama down / API error → cli-finding-classifier 内 fallback で human_review + fallback_reason、report に出力
JSON parse 失敗 → markdown に raw 含めて記録

smoke dogfood 結果

868 行の現実 PR diff (Phase c 自身) を流したところ:

{"lint_findings":[],"screen_decision":"human_review","fallback_reason":"ollama error: parse: JSON parse error: missing field `screen_decision`"}

= mistral:7b は大規模 diff で JSON 構造を欠落させがちと判明。fallback path が graceful に処理。Phase d (PR-based dogfood) で頻度・対策方針を観測。

副次対応

step_timeout = 120 → 180 (Phase b' end-to-end test の cargo test --ignored が境界 117s で timeout していた、commit 73903d7 で実証)

Test plan

cargo test -p cli-push-runner — 47 件 pass
cargo test -p cli-finding-classifier --test lint_screen_evals -- --ignored — 12 件 + 1 ignored 完走
cargo clippy / fmt clean
smoke: cli-finding-classifier に Phase c 自身の diff (868 行) を流して fallback path 確認
実 PR push で takt simplicity-review が .takt/lint-screen-report.md を context として認識するか (Phase d で観測)

Out of scope (PR-B 行き)

Phase c 完了マーク (analysis.md)
post-merge-feedback (PR feat(cli-finding-classifier): Phase b' lint-screen evals — N=12 + multi-metrics + prompt v3 (83.3% GO) #131) 採用 follow-up
- T2-1: fixture metadata consistency test
- T3-1: ADR-038 Fixture 作成時チェックリスト
Phase d (PR-based 実環境 dogfood)

Summary by CodeRabbit

リリースノート

新機能
- オプションのlint-screenプリプッシュフェーズを追加（デフォルトで無効）。外部分析を実行し、マークダウンレポートを生成します。
ドキュメント
- コードレビュー中に補足的なlintレポートを活用するガイダンスを追加しました。
Chores
- 品質ゲートのタイムアウトを120秒から180秒に延長しました。
- lint-screen設定オプションを追加しました。

coderabbitai · 2026-05-08T19:01:28Z

Warning

Rate limit exceeded

@aloekun has exceeded the limit for the number of commits that can be reviewed per hour. Please wait 41 minutes and 8 seconds before requesting another review.

You’ve run out of usage credits. Purchase more in the billing tab.

⌛ How to resolve this issue?

After the wait time has elapsed, a review can be triggered using the @coderabbitai review command as a PR comment. Alternatively, push new commits to this PR.

We recommend that you space out your commits to avoid hitting the rate limit.

🚦 How do rate limits work?

CodeRabbit enforces hourly rate limits for each developer per organization.

Our paid plans have higher rate limits than the trial, open-source and free plans. In all cases, we re-allow further reviews after a brief timeout.

Please see our FAQ for further information.

ℹ️ Review info

⚙️ Run configuration

Configuration used: Organization UI

Review profile: CHILL

Plan: Pro

Run ID: 0df2e926-dc48-4086-a55b-69d8d22f6b97

📥 Commits

Reviewing files that changed from the base of the PR and between 73903d7 and 8416d3e.

⛔ Files ignored due to path filters (1)

Cargo.lock is excluded by !**/*.lock

📒 Files selected for processing (7)

.takt/facets/instructions/review-simplicity.md
push-runner-config.toml
src/cli-push-runner/Cargo.toml
src/cli-push-runner/src/config.rs
src/cli-push-runner/src/main.rs
src/cli-push-runner/src/stages/lint_screen.rs
src/cli-push-runner/src/stages/mod.rs

📝 Walkthrough

概要

PR は Diff に対して外部 lint-screen 分類器を実行し、マークダウン形式のレポートを生成する新しいパイプラインステージを追加します。このステージは設定で無効化でき、push パイプラインをブロックしません。

変更内容

Lint-Screen Feature

レイヤー / ファイル	概要
設定構造とコントラクト `src/cli-push-runner/src/config.rs`, `push-runner-config.toml`, `src/cli-push-runner/Cargo.toml`	`LintScreenConfig` 構造体（enabled、exe_path、model、endpoint、timeout_secs、max_diff_lines、output_path フィールドを含む）、デフォルト定数（タイムアウト 60 秒、最大 diff 行数 5000、モデル mistral:7b、Ollama エンドポイント、分類器実行ファイルパス、出力ファイル `.takt/lint-screen-report.md`）、トップレベル `Config` への統合。quality_gate step_timeout を 120 から 180 に増加。serde_json 依存関係を追加。
ステージ実装 `src/cli-push-runner/src/stages/lint_screen.rs`	run_lint_screen オーケストレータ、read_diff（最大行数の検証）、invoke_classifier（外部プロセス管理とタイムアウト処理）、write_report（マークダウン出力）、format_report（JSON パース）、render_findings_table（マークダウンテーブル生成）、sanitize_cell（セルエスケープ）。エラーは skip 理由として記録され、パイプラインをブロックしません。
パイプライン統合 `src/cli-push-runner/src/main.rs`, `src/cli-push-runner/src/stages/mod.rs`	run_diff_and_lint_screen ヘルパー（diff 実行と条件付き lint-screen 呼び出しをカプセル化）、run_pipeline の refactor（新しいヘルパー呼び出し）、lint_screen サブモジュール登録と再エクスポート。
テストと文書化 `src/cli-push-runner/src/config.rs`, `src/cli-push-runner/src/stages/lint_screen.rs`, `.takt/facets/instructions/review-simplicity.md`	Config テスト更新（lint_screen: None を含む）、lint_screen ユニットテスト（JSON 解析エラー復旧、空の findings、セル sanitization、read_diff エラーケース）、レビュー簡潔化ガイドに lint-screen レポート使用法を追加。

推定コードレビュー難度

🎯 3 (中程度) | ⏱️ ~20 分

関連する可能性のある PR

aloekun/claude-code-hook-test#130: 当該 PR が lint-screen の実行と markdown レポート生成を実装する一方で、こちらは lint-screen モード、API（screen_diff）、プロンプト、およびそのフィクスチャ/JSON 出力を追加するため、直接関連します。
aloekun/claude-code-hook-test#47: 同じレビュー文書（.takt/facets/instructions/review-simplicity.md）を変更するため直接関連します。
aloekun/claude-code-hook-test#120: 設定フィールド（model/endpoint/timeout）、デフォルト無効動作、JSON I/O を含む同じパターンの設定可能なサブプロセスベースの分類器統合を異なるコンポーネントで実装するため関連します。

🚥 Pre-merge checks | ✅ 4 | ❌ 1

❌ Failed checks (1 warning)

Check name	Status	Explanation	Resolution
Docstring Coverage	⚠️ Warning	Docstring coverage is 60.87% which is insufficient. The required threshold is 80.00%.	Write docstrings for the functions missing them to satisfy the coverage threshold.

✅ Passed checks (4 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	PR title clearly and specifically describes the main change: adding a Phase C MVP lint-screen pre-push step to cli-push-runner with default OFF configuration, which aligns directly with the changeset contents.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.

_{✏️ Tip: You can configure your own custom pre-merge checks in the settings.}

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

coderabbitai

Actionable comments posted: 1

🧹 Nitpick comments (1)

src/cli-push-runner/src/config.rs (1)

113-227: ⚡ Quick win

[lint_screen] セクションのパース動作を検証するテストが不在

config_parses_with_diff が [diff] セクションのパースを検証しているように、[lint_screen] セクション全フィールドのパースをテストするケースがない。serde の field 名と TOML キー名は完全一致が必要なため、フィールドのリネームや追加が silent な None へのフォールバックに化けるリスクを防ぐためにも追加が望ましい。

🧪 テスト追加の例

+    #[test]
+    fn config_parses_lint_screen_section() {
+        let toml_str = r#"
+[quality_gate]
+[[quality_gate.groups]]
+name = "test"
+commands = ["echo ok"]
+
+[takt]
+workflow = "w"
+task = "t"
+
+[push]
+command = "echo push"
+
+[lint_screen]
+enabled = true
+exe_path = "custom/classifier.exe"
+model = "mistral:7b"
+endpoint = "http://localhost:11434"
+timeout_secs = 30
+max_diff_lines = 1000
+output_path = ".takt/custom-report.md"
+"#;
+        let config: Config = toml::from_str(toml_str).unwrap();
+        let ls = config.lint_screen.unwrap();
+        assert!(ls.enabled);
+        assert_eq!(ls.exe_path.as_deref(), Some("custom/classifier.exe"));
+        assert_eq!(ls.model.as_deref(), Some("mistral:7b"));
+        assert_eq!(ls.endpoint.as_deref(), Some("http://localhost:11434"));
+        assert_eq!(ls.timeout_secs, Some(30));
+        assert_eq!(ls.max_diff_lines, Some(1000));
+        assert_eq!(ls.output_path.as_deref(), Some(".takt/custom-report.md"));
+    }

🤖 Prompt for AI Agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

In `@src/cli-push-runner/src/config.rs` around lines 113 - 227, Add a new unit
test (similar to config_parses_with_diff) named e.g. config_parses_lint_screen
that deserializes a TOML string containing a full [lint_screen] section and
asserts every field on Config.lint_screen (or the LintScreen type) is
Some/equals the expected values; locate the Config struct and existing tests
(e.g. config_parses_with_diff, config_parses_full_without_diff) to mirror style,
ensure you check optional fields unwrap_or defaults where applicable and assert
that no field silently remains None due to a naming mismatch.

🤖 Prompt for all review comments with AI agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

Inline comments:
In `@src/cli-push-runner/src/stages/lint_screen.rs`:
- Around line 199-216: The table-building code in render_findings_table leaves s
(severity), r (rule), file and line unescaped which can break the Markdown
table; wrap each of those values with sanitize_cell before formatting (i.e. pass
s, r, file and the computed line string through sanitize_cell) and ensure the
line value is produced as a deterministic string (use the existing
map/unwrap_or_else flow but sanitize its result) so all four fields are escaped
consistently alongside issue and suggestion.

---

Nitpick comments:
In `@src/cli-push-runner/src/config.rs`:
- Around line 113-227: Add a new unit test (similar to config_parses_with_diff)
named e.g. config_parses_lint_screen that deserializes a TOML string containing
a full [lint_screen] section and asserts every field on Config.lint_screen (or
the LintScreen type) is Some/equals the expected values; locate the Config
struct and existing tests (e.g. config_parses_with_diff,
config_parses_full_without_diff) to mirror style, ensure you check optional
fields unwrap_or defaults where applicable and assert that no field silently
remains None due to a naming mismatch.

🪄 Autofix (Beta)

Fix all unresolved CodeRabbit comments on this PR:

Push a commit to this branch (recommended)
Create a new PR with the fixes

ℹ️ Review info

⚙️ Run configuration

Configuration used: Organization UI

Review profile: CHILL

Plan: Pro

Run ID: c4d07d95-a1ce-45dd-92db-be62e6706fff

📥 Commits

Reviewing files that changed from the base of the PR and between fb33818 and 73903d7.

⛔ Files ignored due to path filters (1)

Cargo.lock is excluded by !**/*.lock

📒 Files selected for processing (7)

.takt/facets/instructions/review-simplicity.md
push-runner-config.toml
src/cli-push-runner/Cargo.toml
src/cli-push-runner/src/config.rs
src/cli-push-runner/src/main.rs
src/cli-push-runner/src/stages/lint_screen.rs
src/cli-push-runner/src/stages/mod.rs

…lt OFF, gating なし、fallback graceful)

…#134) - todo6.md / todo-summary.md に Bundle j 3 タスク追加 (順位 94-96) - 94: docs/ 内 ../docs/ 相対パストラップ検出 lint rule (Tier 1, S) - 95: docs/todo*.md preamble file count 自動照合 (Tier 2, S) - 96: Markdown cross-reference validator CI step (Tier 2, M) - local-llm-offload-analysis.md に Phase c MVP 完了 (PR #132) を反映 - 実装方針変更 (takt facet -> cli-push-runner stage) を §1 Phase c に記述 - Phase c+ (Bundle i) として PR #132 post-merge-feedback 採用 3 件 - §8.E を MVP land 済みステータスに更新 - 再開チェックリストに Phase c smoke / Bundle i 着手手順を追加 - todo-summary.md retrospective に Bundle j (PR #133 post-merge-feedback) 追記

…必須 follow-up (順位 91 + 92 + 93) (#135) * feat(cli-finding-classifier, cli-push-runner): Bundle i — Phase d 着手前必須 follow-up (順位 91 + 92 + 93) PR #132 (Phase c MVP land) の post-merge-feedback で採用された 3 件を 1 PR にまとめて land。 ## 順位 91 (Tier 2 #4): [lint_screen] config parse test - src/cli-push-runner/src/config.rs に 5 tests を追加 - silent field rename / 追加で None fallback する failure mode を unit test で防止 - full fields / minimal only enabled / absent yields None / numeric defaults / string defaults の 5 軸独立検証 ## 順位 92 (Tier 2 #5): scale-aware eval fixtures (200+ 行) - eval13-large-refactor-real.diff (5 file / 280 行) — context 限界 + JSON 完全性 - eval14-mid-mixed.diff (3 file / 153 行) — mid-scale recall 安定性 - eval15-syntax-stress.diff (1 file / 208 行) — 単 file 長尺の schema 完全性 - lint-screen-evals.json に id 13/14/15 baseline (auto_fix lane × 13 findings 合計) 追加 - count test を rename + 上限緩和 (eval_set_loads_and_has_at_least_phase_b_prime_baseline_count) - Bundle i 実体スモーク test (eval_set_includes_bundle_i_scale_aware_fixtures) 追加 ### dogfood 結果 (mistral:7b / temperature=0) agreement = 11/15 = 73.3% (Phase b' 75% から marginal 劣化 = fixture が設計通り failure mode を再現) eval13 (280 行): JSON parse error 'missing field screen_decision' → fallback path 作動 = PR #132 smoke (868 行 diff) で観測した failure mode を decisive に再現 eval15 (208 行): JSON parse error 'missing field severity at line 38' = nested field omission の別 failure mode を新規捕捉 eval14 (153 行): JSON 完全だが recall 33% (3 baseline 中 1 件のみ TP) aggregate precision=76.2% recall=51.6% latency p50=4591ms p95=8370ms verdict CONDITIONAL-GO agreement < 75% 未達理由は eval13/15 の fallback (= fixture が設計通り作動した結果) で mechanical に説明可能。Phase d 投入前の必須 measurement を取得 (todo6.md L164 「未達理由が文書化される」branch を満たす)。§8.D v4 prompt 改訂は別 bundle に切り出し。 ## 順位 93 (Tier 3 #8): coding-style.md partial fix anti-pattern codify - ~/.claude/rules/common/coding-style.md § Cross-File Reference Lifecycle に「変更差分外への partial fix 再発」anti-pattern を追加 - PR #94 / #111 / #132 を inline cite (実証ベース) - family_tag を grep -rn で全 path 検索する対処手順、partial fix の意図的切り出しを明記 ## Phase d 着手の前提条件 update Bundle i land で以下が揃った: - (a) [lint_screen] config silent failure 防止 (順位 91) - (b) scale-aware fixtures による failure mode の reproducible measurement (順位 92) - (c) cross-file partial fix anti-pattern の global rule 化 (順位 93) 次は §8.D v4 prompt 改訂で大規模 diff の JSON 完全性を改善するループ (Phase d 着手前の最終 gate)。 * fix(cli-finding-classifier): CodeRabbit Major #r3213115045 — eval count 下限を Bundle i baseline 15 に固定 >=12 だと既存 fixture 削除を検出できないため >= 15 に変更し regression 防止。将来の fixture 追加 (>15) は許容。

coderabbitai Bot reviewed May 8, 2026

View reviewed changes

Comment thread src/cli-push-runner/src/stages/lint_screen.rs

feat(cli-push-runner): Phase c MVP — pre-push lint-screen step (defau…

8416d3e

…lt OFF, gating なし、fallback graceful)

aloekun force-pushed the feature/local-llm-phase-c-mvp branch from 73903d7 to 8416d3e Compare May 8, 2026 19:20

aloekun merged commit fd54b08 into master May 8, 2026
1 check passed

aloekun deleted the feature/local-llm-phase-c-mvp branch May 8, 2026 20:18

aloekun mentioned this pull request May 9, 2026

docs: Bundle j 登録 + Phase c MVP land 反映 (PR #132 post-merge-feedback) #134

Merged

4 tasks

aloekun mentioned this pull request May 9, 2026

feat(cli-finding-classifier, cli-push-runner): Bundle i — Phase d 着手前必須 follow-up (順位 91 + 92 + 93) #135

Merged

5 tasks

This was referenced May 10, 2026

docs: Phase d kickoff prep + 順位 98 追加 + dogfood roster 反映 (3 commits) #138

Merged

feat(lib-ollama-client): DEFAULT_NUM_CTX → 32768 (Phase C / 順位 98 root cause fix) #143

Merged

This was referenced May 11, 2026

feat(cli-push-runner): lint_screen stderr → report 取込 (Phase D 前提整備、順位 109) #144

Merged

feat(cli-push-runner): LINT_SCREEN_ENABLED env var override (順位 115) #147

Merged

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

feat(cli-push-runner): Phase c MVP — pre-push lint-screen step (default OFF)#132

feat(cli-push-runner): Phase c MVP — pre-push lint-screen step (default OFF)#132
aloekun merged 1 commit into
masterfrom
feature/local-llm-phase-c-mvp

aloekun commented May 8, 2026 •

edited by coderabbitai Bot

Loading

Uh oh!

coderabbitai Bot commented May 8, 2026 •

edited

Loading

Rate limit exceeded

概要

変更内容

推定コードレビュー難度

関連する可能性のある PR

❌ Failed checks (1 warning)

Uh oh!

coderabbitai Bot left a comment

Uh oh!

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

Conversation

aloekun commented May 8, 2026 • edited by coderabbitai Bot Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Summary

アーキテクチャ

fallback 設計 (push を block しない)

smoke dogfood 結果

副次対応

Test plan

Out of scope (PR-B 行き)

Summary by CodeRabbit

リリースノート

Uh oh!

coderabbitai Bot commented May 8, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Rate limit exceeded

概要

変更内容

推定コードレビュー難度

関連する可能性のある PR

❌ Failed checks (1 warning)

Uh oh!

coderabbitai Bot left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

aloekun commented May 8, 2026 •

edited by coderabbitai Bot

Loading

coderabbitai Bot commented May 8, 2026 •

edited

Loading