Repository navigation
Claude/context aware review pl gx t - #22
Merged
Merged
Conversation
Add context-aware review detection to /harness-review: - Auto-detects review type based on recent activity - After /plan-with-agent → Plan Review - After /work → Code Review - After adding tasks → Scope Review Add 8 new expert files for plan and scope reviews: - Plan: clarity-expert, feasibility-expert, dependencies-expert, acceptance-expert - Scope: scope-creep-expert, priority-expert, scope-feasibility-expert, impact-expert Update codex-parallel-review.md to document 3 review types with 4 experts each. Unified expert count between Codex mode ON and OFF.
|
Warning Rate limit exceeded
⌛ How to resolve this issue?After the wait time has elapsed, a review can be triggered using the 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. Walkthroughマニュアル中心の大規模更新で、スキル名を Changes
Sequence Diagram(s)sequenceDiagram
participant User as ユーザー
participant Harness as harness-review (Orchestrator)
participant Detector as Context Detector
participant Codex as Codex / MCP
participant Expert1 as Expert 1
participant Expert2 as Expert 2
participant Expert3 as Expert 3
participant Expert4 as Expert 4
participant Aggregator as Aggregator
participant CommitGuard as Commit Guard / State
User->>Harness: harness-review 実行 (files, context_from?, review_mode?)
Harness->>Detector: 最近の活動を解析（自動判定）
Detector-->>Harness: review_type (code | plan | scope)
Harness->>Codex: 該当 review_type のプロンプトをロード
Codex->>Expert1: 並列呼び出し (prompt A)
Codex->>Expert2: 並列呼び出し (prompt B)
Codex->>Expert3: 並列呼び出し (prompt C)
Codex->>Expert4: 並列呼び出し (prompt D)
Expert1-->>Aggregator: 評価・改善案
Expert2-->>Aggregator: 評価・改善案
Expert3-->>Aggregator: 評価・改善案
Expert4-->>Aggregator: 評価・改善案
Aggregator->>Harness: 集約結果（総評・視点別）
Harness->>CommitGuard: 判定要求（APPROVE / REQUEST_CHANGES）
CommitGuard-->>User: commit 許可/ブロック、および状態ファイル出力
Harness-->>User: 最終レポート（総評・視点別評価・改善案）
Estimated code review effort🎯 3 (Moderate) | ⏱️ ~20 minutes Possibly related PRs
Poem
🚥 Pre-merge checks | ✅ 2 | ❌ 1❌ Failed checks (1 inconclusive)
✅ Passed checks (2 passed)
✏️ 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. Comment |
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
There was a problem hiding this comment.
Actionable comments posted: 7
Caution
Some comments are outside the diff and can’t be posted inline due to platform limitations.
⚠️ Outside diff range comments (6)
opencode/skills/codex-review/references/codex-review-integration.md (1)
26-33: 「review スキル起動」表記を rename 後に合わせて修正
/harness-reviewへ統一した文脈なので「review スキル起動」は不整合です。「harness-review スキル起動」へ置換してください。📝 変更提案
- review スキル起動 + harness-review スキル起動opencode/skills/harness-review/SKILL.md (1)
1-6: SSOT ファイル（decisions.md / patterns.md）が存在しませんコーディングガイドラインで「opencode/**/*.md 変更時は
.claude/memory/decisions.mdと.claude/memory/patterns.mdに決定事項と再利用パターンを記録する」と規定されていますが、両ファイルが存在しません。
.claude/memory/ディレクトリ内に以下を作成し、SKILL.md の変更内容（スキル名「review」→「harness-review」への変更理由とパターン）を記録してください：
.claude/memory/decisions.md- 名前変更の決定理由（Why）.claude/memory/patterns.md- スキル設計の再利用可能パターン（How）opencode/skills/codex-review/references/codex-parallel-review.md (2)
33-52: 「4エキスパート化」と「8回呼び出し」の記述が矛盾しています。
冒頭で4エキスパートに整理したのに、このセクションは旧ルールのままで誤誘導になります。🛠️ 修正案（4エキスパート前提 or 動的記述）
-| ✅ 各エキスパートを **個別の MCP 呼び出し** で実行 | `mcp__codex__codex` を8回呼び出し | +| ✅ 各エキスパートを **個別の MCP 呼び出し** で実行 | `mcp__codex__codex` を有効エキスパート数だけ呼び出し | ... -| ✅ **1つのレスポンス内で8つの MCP 呼び出しを並列実行** | Claude の並列ツール呼び出し機能を使用 | +| ✅ **1つのレスポンス内で有効エキスパート数の MCP 呼び出しを並列実行** | Claude の並列ツール呼び出し機能を使用 | ... -> **理由**: 8エキスパート並列でも 1500文字×8 = 12,000文字 ≒ 4,000トークン程度で収まる +> **理由**: 有効エキスパート数 × 1500文字 で概ねトークン上限に収まるよう設計Also applies to: 207-218
161-172: 4エキスパート運用の説明に、6〜8エキスパートの例が混在しています。
レビュータイプ定義と例・集計表の前提がズレているため、運用手順の整合性が崩れています。🛠️ 修正案（例と集計表を4エキスパートに合わせる）
-例1: Webフロントエンドでコード変更あり -→ Security, Accessibility, Performance, Quality, SEO, Architect -→ 6エキスパート並列（Plan Reviewer, Scope Analyst は除外） +例1: Webフロントエンドでコード変更あり +→ Security, Accessibility, Performance, Quality +→ 4エキスパート並列 -例2: CLI プラグインでドキュメントのみ変更 -→ Quality, Architect, Plan Reviewer, Scope Analyst -→ 4エキスパート並列（Security, Performance, Accessibility, SEO は除外） +例2: CLI プラグインでドキュメントのみ変更 +→ Clarity, Feasibility, Dependencies, Acceptance +→ 4エキスパート並列-| Security | B | 0 | 1 | 2 | 3 | -| Accessibility | A | 0 | 0 | 1 | 2 | -| Performance | C | 0 | 2 | 3 | 1 | -| Quality | B | 0 | 0 | 4 | 5 | -| SEO | A | 0 | 0 | 0 | 2 | -| Architect | B | 0 | 1 | 1 | 0 | -| Plan Reviewer | APPROVE | - | - | - | - | -| Scope Analyst | Proceed | - | - | - | - | +| Security | B | 0 | 1 | 2 | 3 | +| Performance | C | 0 | 2 | 3 | 1 | +| Quality | B | 0 | 0 | 4 | 5 | +| Accessibility | A | 0 | 0 | 1 | 2 |-| `experts/seo-expert.md` | SEO エキスパートプロンプト | -| `experts/architect-expert.md` | 設計エキスパートプロンプト | -| `experts/plan-reviewer-expert.md` | 計画レビューエキスパートプロンプト | -| `experts/scope-analyst-expert.md` | 要件分析エキスパートプロンプト | +| `experts/accessibility-expert.md` | a11y エキスパートプロンプト |Also applies to: 224-239, 324-336
skills/codex-review/references/codex-parallel-review.md (1)
47-52: 「8回/8エキスパート」の記述が4エキスパート方針と矛盾しています。運用指示が誤ると不要な並列呼び出し・コスト増につながるため、現在の4エキスパートモデルに合わせて文言を更新してください。
🔧 修正案
-| ✅ 各エキスパートを **個別の MCP 呼び出し** で実行 | `mcp__codex__codex` を8回呼び出し | +| ✅ 各エキスパートを **個別の MCP 呼び出し** で実行 | `mcp__codex__codex` を選択されたエキスパート数だけ呼び出し | ... -| ✅ **1つのレスポンス内で8つの MCP 呼び出しを並列実行** | Claude の並列ツール呼び出し機能を使用 | +| ✅ **1つのレスポンス内で4つの MCP 呼び出しを並列実行** | Claude の並列ツール呼び出し機能を使用 | ... -> **理由**: 8エキスパート並列でも 1500文字×8 = 12,000文字 ≒ 4,000トークン程度で収まる +> **理由**: 4エキスパート並列でも 1500文字×4 = 6,000文字 ≒ 2,000トークン程度で収まるAlso applies to: 218-218
opencode/commands/core/harness-review.md (1)
1-510: SSOT（decisions/patterns）への追記が必要です。
opencode/**/*.md 変更時は.claude/memory/decisions.mdと.claude/memory/patterns.mdへの記録が必須です。更新が見当たらないため追加してください。As per coding guidelines.
🤖 Fix all issues with AI agents
In `@commands/core/harness-review.md`:
- Around line 70-74: 該当のフェンスコードブロック（中身が "Use Skill tool:" と `skill:
"claude-code-harness:harness-review"`
を含む箇所）に言語指定が欠けているため、開始バックティックに言語識別子を追加して例えば ```text のように変更してください（つまり ``` を
```text に置き換える） so the fenced code block has an explicit language identifier.
In `@opencode/commands/core/harness-review.md`:
- Around line 70-74: The fenced code block in harness-review.md is missing a
language specifier; update the snippet that currently starts with ``` to include
a language (e.g., change the opening fence to ```text) so markdownlint stops
flagging it; search for the block containing `Use Skill tool:` (the snippet used
to show how to call the skill) and replace the opening fence with a
language-tagged fence like ```text while keeping the block content unchanged.
In `@opencode/skills/codex-review/references/experts/dependencies-expert.md`:
- Around line 36-42: The fenced code block under "Dependency Map" lacks a
language tag causing MD040; update the triple-backtick fence that contains the
ASCII diagram (the block showing "[Task A] ──→ [Task B] ──→ [Task C]" with
"[Task D]") to include a language specifier (e.g., use ```text) so the block
becomes ```text ... ``` to satisfy markdownlint.
In `@opencode/skills/codex-review/references/experts/priority-expert.md`:
- Around line 45-49: Fix the heading typo by replacing the text "Misprioritzed
Items" with "Misprioritized Items" in the markdown section (look for the heading
line "#### Misprioritzed Items" under "### Priority Issues"); update only that
heading to correct the spelling and keep the rest of the list content unchanged.
In `@skills/codex-review/references/codex-parallel-review.md`:
- Around line 17-29: The fenced code block containing the flow diagram (starting
with "Claude (オーケストレーター)" and ending with "結果統合 → 判定") needs a language tag to
satisfy markdownlint MD040; update the opening fence from ``` to ```text so the
block becomes a ```text fenced code block, preserving the diagram contents
exactly.
In `@skills/codex-review/references/experts/priority-expert.md`:
- Around line 45-49: Fix the spelling mistake in the subsection heading "####
Misprioritzed Items" under "### Priority Issues": change "Misprioritzed" to
"Misprioritized" so the heading reads "#### Misprioritized Items"; update any
other occurrences of the misspelled word in this file
(skills/codex-review/references/experts/priority-expert.md) to maintain
consistency.
In `@templates/rules/skills-gate.md.template`:
- Line 26: 下部の例示ブロックで旧表記の "review" が残っているため、テンプレート内のすべての呼び出し例を統一して
"harness-review" に置換してください（例にある行「コードをレビューする | `harness-review`」と、例示ブロック内の
`review` 呼び出しをすべて `harness-review` に変更し、一貫性が取れていることを確認してください）。
🧹 Nitpick comments (1)
skills/codex-review/references/codex-review-integration.md (1)
190-191: 本文内の呼称も “harness-review” に統一することを推奨します。
図中の「review スキル起動」や YAML 例のreview:が残っている場合は、今回のリネーム方針と合わせて更新すると一貫性が上がります。
Comment on lines
70
to
74
| **How to call**: | ||
| ``` | ||
| Use Skill tool: | ||
| skill: "claude-code-harness:review" | ||
| skill: "claude-code-harness:harness-review" | ||
| ``` |
There was a problem hiding this comment.
フェンスコードに言語指定がありません。
markdownlint の指摘どおり、言語を指定してください。
✏️ 修正案
-```
+```text
Use Skill tool:
skill: "claude-code-harness:harness-review"</details>
<!-- suggestion_start -->
<details>
<summary>📝 Committable suggestion</summary>
> ‼️ **IMPORTANT**
> Carefully review the code before committing. Ensure that it accurately replaces the highlighted code, contains no missing lines, and has no issues with indentation. Thoroughly test & benchmark the code to ensure it meets the requirements.
```suggestion
**How to call**:
🧰 Tools
🪛 markdownlint-cli2 (0.18.1)
71-71: Fenced code blocks should have a language specified
(MD040, fenced-code-language)
🤖 Prompt for AI Agents
In `@commands/core/harness-review.md` around lines 70 - 74, 該当のフェンスコードブロック（中身が
"Use Skill tool:" と `skill: "claude-code-harness:harness-review"`
を含む箇所）に言語指定が欠けているため、開始バックティックに言語識別子を追加して例えば ```text のように変更してください（つまり ``` を
```text に置き換える） so the fenced code block has an explicit language identifier.
Comment on lines
70
to
74
| **How to call**: | ||
| ``` | ||
| Use Skill tool: | ||
| skill: "claude-code-harness:review" | ||
| skill: "claude-code-harness:harness-review" | ||
| ``` |
There was a problem hiding this comment.
フェンスコードに言語指定がありません。
markdownlint の指摘どおり、言語を指定してください。
✏️ 修正案
-```
+```text
Use Skill tool:
skill: "claude-code-harness:harness-review"</details>
<details>
<summary>🧰 Tools</summary>
<details>
<summary>🪛 markdownlint-cli2 (0.18.1)</summary>
71-71: Fenced code blocks should have a language specified
(MD040, fenced-code-language)
</details>
</details>
<details>
<summary>🤖 Prompt for AI Agents</summary>
In @opencode/commands/core/harness-review.md around lines 70 - 74, The fenced
code block in harness-review.md is missing a language specifier; update the
snippet that currently starts with to include a language (e.g., change the opening fence totext) so markdownlint stops flagging it; search for the
block containing Use Skill tool: (the snippet used to show how to call the
skill) and replace the opening fence with a language-tagged fence like ```text
while keeping the block content unchanged.
</details>
<!-- fingerprinting:phantom:triton:eagle -->
<!-- This is an auto-generated comment by CodeRabbit -->
Comment on lines
+36
to
+42
| ### Dependency Map | ||
|
|
||
| ``` | ||
| [Task A] ──→ [Task B] ──→ [Task C] | ||
| │ ↑ | ||
| └──→ [Task D] ──────────┘ | ||
| ``` |
There was a problem hiding this comment.
フェンスコードに言語指定がありません。
markdownlint の MD040 回避のため、言語指定（例: text）を付けてください。
🔧 修正案
-```
+```text
[Task A] ──→ [Task B] ──→ [Task C]
│ ↑
└──→ [Task D] ──────────┘</details>
<!-- suggestion_start -->
<details>
<summary>📝 Committable suggestion</summary>
> ‼️ **IMPORTANT**
> Carefully review the code before committing. Ensure that it accurately replaces the highlighted code, contains no missing lines, and has no issues with indentation. Thoroughly test & benchmark the code to ensure it meets the requirements.
```suggestion
### Dependency Map
🧰 Tools
🪛 markdownlint-cli2 (0.18.1)
42-42: Fenced code blocks should have a language specified
(MD040, fenced-code-language)
🤖 Prompt for AI Agents
In `@opencode/skills/codex-review/references/experts/dependencies-expert.md`
around lines 36 - 42, The fenced code block under "Dependency Map" lacks a
language tag causing MD040; update the triple-backtick fence that contains the
ASCII diagram (the block showing "[Task A] ──→ [Task B] ──→ [Task C]" with
"[Task D]") to include a language specifier (e.g., use ```text) so the block
becomes ```text ... ``` to satisfy markdownlint.
Comment on lines
+45
to
+49
| ### Priority Issues | ||
|
|
||
| #### Misprioritzed Items | ||
| - [ ] [Task] - Currently P[X], should be P[Y] | ||
| - Reason: [why priority is wrong] |
There was a problem hiding this comment.
見出しのスペルミスを修正してください。
「Misprioritzed」は「Misprioritized」です。
🔧 修正案
-#### Misprioritzed Items
+#### Misprioritized Items📝 Committable suggestion
‼️ IMPORTANT
Carefully review the code before committing. Ensure that it accurately replaces the highlighted code, contains no missing lines, and has no issues with indentation. Thoroughly test & benchmark the code to ensure it meets the requirements.
Suggested change
| ### Priority Issues | |
| #### Misprioritzed Items | |
| - [ ] [Task] - Currently P[X], should be P[Y] | |
| - Reason: [why priority is wrong] | |
| ### Priority Issues | |
| #### Misprioritized Items | |
| - [ ] [Task] - Currently P[X], should be P[Y] | |
| - Reason: [why priority is wrong] |
🤖 Prompt for AI Agents
In `@opencode/skills/codex-review/references/experts/priority-expert.md` around
lines 45 - 49, Fix the heading typo by replacing the text "Misprioritzed Items"
with "Misprioritized Items" in the markdown section (look for the heading line
"#### Misprioritzed Items" under "### Priority Issues"); update only that
heading to correct the spelling and keep the rest of the list content unchanged.
Comment on lines
17
to
29
| ``` | ||
| Claude (オーケストレーター) | ||
| ↓ | ||
| 並列 MCP 呼び出し | ||
| ├── Security Expert | ||
| ├── Accessibility Expert | ||
| ├── Performance Expert | ||
| ├── Quality Expert | ||
| ├── SEO Expert | ||
| ├── Architect Expert | ||
| ├── Plan Reviewer Expert | ||
| └── Scope Analyst Expert | ||
| レビュータイプ判定 | ||
| ↓ | ||
| 並列 MCP 呼び出し (4 experts) | ||
| ├── Expert 1 | ||
| ├── Expert 2 | ||
| ├── Expert 3 | ||
| └── Expert 4 | ||
| ↓ | ||
| 結果統合 → コミット判定 | ||
| 結果統合 → 判定 | ||
| ``` |
There was a problem hiding this comment.
フェンスコードに言語指定を付与してください。
markdownlint MD040 対応のため、text などを指定するのが無難です。
🔧 修正案
-```
+```text
Claude (オーケストレーター)
↓
レビュータイプ判定
↓
並列 MCP 呼び出し (4 experts)
├── Expert 1
├── Expert 2
├── Expert 3
└── Expert 4
↓
結果統合 → 判定</details>
<details>
<summary>🧰 Tools</summary>
<details>
<summary>🪛 markdownlint-cli2 (0.18.1)</summary>
17-17: Fenced code blocks should have a language specified
(MD040, fenced-code-language)
</details>
</details>
<details>
<summary>🤖 Prompt for AI Agents</summary>
In @skills/codex-review/references/codex-parallel-review.md around lines 17 -
29, The fenced code block containing the flow diagram (starting with "Claude
(オーケストレーター)" and ending with "結果統合 → 判定") needs a language tag to satisfy
markdownlint MD040; update the opening fence from totext so the block
becomes a ```text fenced code block, preserving the diagram contents exactly.
</details>
<!-- fingerprinting:phantom:triton:eagle -->
<!-- This is an auto-generated comment by CodeRabbit -->
Comment on lines
+45
to
+49
| ### Priority Issues | ||
|
|
||
| #### Misprioritzed Items | ||
| - [ ] [Task] - Currently P[X], should be P[Y] | ||
| - Reason: [why priority is wrong] |
There was a problem hiding this comment.
見出しのスペルミスを修正してください。
“Misprioritzed” → “Misprioritized”。
✏️ 修正案
-#### Misprioritzed Items
+#### Misprioritized Items📝 Committable suggestion
‼️ IMPORTANT
Carefully review the code before committing. Ensure that it accurately replaces the highlighted code, contains no missing lines, and has no issues with indentation. Thoroughly test & benchmark the code to ensure it meets the requirements.
Suggested change
| ### Priority Issues | |
| #### Misprioritzed Items | |
| - [ ] [Task] - Currently P[X], should be P[Y] | |
| - Reason: [why priority is wrong] | |
| ### Priority Issues | |
| #### Misprioritized Items | |
| - [ ] [Task] - Currently P[X], should be P[Y] | |
| - Reason: [why priority is wrong] |
🤖 Prompt for AI Agents
In `@skills/codex-review/references/experts/priority-expert.md` around lines 45 -
49, Fix the spelling mistake in the subsection heading "#### Misprioritzed
Items" under "### Priority Issues": change "Misprioritzed" to "Misprioritized"
so the heading reads "#### Misprioritized Items"; update any other occurrences
of the misspelled word in this file
(skills/codex-review/references/experts/priority-expert.md) to maintain
consistency.
| |------|--------|-----------| | ||
| | 機能を実装する | `impl` | Skill ツールで「impl」を呼び出す | | ||
| | コードをレビューする | `review` | Skill ツールで「review」を呼び出す | | ||
| | コードをレビューする | `harness-review` | Skill ツールで「harness-review」を呼び出す | |
There was a problem hiding this comment.
例示の “review” 表記も “harness-review” に統一してください。
下部のブロック例（例: “review” 呼び出し）が旧表記のままなので混乱します。
🤖 Prompt for AI Agents
In `@templates/rules/skills-gate.md.template` at line 26, 下部の例示ブロックで旧表記の "review"
が残っているため、テンプレート内のすべての呼び出し例を統一して "harness-review" に置換してください（例にある行「コードをレビューする |
`harness-review`」と、例示ブロック内の `review` 呼び出しをすべて `harness-review`
に変更し、一貫性が取れていることを確認してください）。
- Update codex-parallel-review.md: clarify "4 experts per review type" instead of misleading "8 experts" references - Add review-type-based config structure (code_experts, plan_experts, scope_experts) with backward compatibility note - Update harness-review SKILL.md with context-aware review type section - Document plan review (clarity, feasibility, dependencies, acceptance) - Document scope review (scope-creep, priority, feasibility, impact) - Sync all changes to opencode/ directory Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
- harness-review SKILL.md: "8 エキスパート" -> "レビュータイプごとに4エキスパート" - harness-review SKILL.md: "8回の MCP 呼び出し" -> "4回の MCP 呼び出し" - codex-parallel-review.md: "8エキスパート並列" -> "4エキスパート並列" - codex-parallel-review.md: Update examples to show review-type-based selection - codex-review SKILL.md: "最大8並列レビュー" -> "4並列レビュー" - codex-review SKILL.md: "6エキスパート有効" -> "Code Reviewの場合、4エキスパート" - Apply all changes to opencode/ directory Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
There was a problem hiding this comment.
Actionable comments posted: 6
Caution
Some comments are outside the diff and can’t be posted inline due to platform limitations.
⚠️ Outside diff range comments (2)
skills/codex-review/references/codex-parallel-review.md (1)
204-218: 設定例が新しい per-type 構成と不整合です。ここでは
config.review.codex.expertsを参照していますが、上部でcode_experts / plan_experts / scope_expertsに移行済みと記載されています。レビュータイプに応じた集合を使う例に更新した方が誤解を防げます。🔧 修正案
-const enabledExperts = Object.entries(config.review.codex.experts) - .filter(([_, enabled]) => enabled) - .map(([name]) => name); +const expertsByType = { + code: config.review.codex.code_experts, + plan: config.review.codex.plan_experts, + scope: config.review.codex.scope_experts, +}; +const enabledExperts = Object.entries(expertsByType[reviewType]) + .filter(([_, enabled]) => enabled) + .map(([name]) => name);opencode/skills/codex-review/references/codex-parallel-review.md (1)
204-218: 設定例が per-type 構成と不整合です。
code_experts / plan_experts / scope_expertsの説明と合うよう、レビュータイプ別の選択例に直すのが安全です。🔧 修正案
-const enabledExperts = Object.entries(config.review.codex.experts) - .filter(([_, enabled]) => enabled) - .map(([name]) => name); +const expertsByType = { + code: config.review.codex.code_experts, + plan: config.review.codex.plan_experts, + scope: config.review.codex.scope_experts, +}; +const enabledExperts = Object.entries(expertsByType[reviewType]) + .filter(([_, enabled]) => enabled) + .map(([name]) => name);
🤖 Fix all issues with AI agents
In `@opencode/skills/codex-review/references/codex-parallel-review.md`:
- Around line 17-29: The fenced code block containing the ASCII flow diagram
(starting with the three backticks and the line "Claude (オーケストレーター)" and ending
with three backticks) is missing a language identifier which triggers MD040; fix
it by adding a language specifier (e.g., change the opening ``` to ```text) so
the block becomes ```text ... ``` and keep the diagram content unchanged to
satisfy the linter.
In `@opencode/skills/codex-review/SKILL.md`:
- Line 28: opencode/**/*.md の変更に対して SSOT
が存在しないため、.claude/memory/decisions.md（Why）と
.claude/memory/patterns.md（How）を作成してください; decisions.md に「なぜ 4
並列レビューに変更したのか」という背景と意思決定根拠（利点・トレードオフ・関係者の合意）を明確に記載し、patterns.md に「4
並列レビューの実行パターン」（エキスパートの役割定義、MCP 呼び出し手順、参照として references/codex-parallel-review.md
へのリンクと実践フロー）を具体的な手順として記載してください; 作成後、opencode/skills/codex-review/SKILL.md
の該当箇所にある「4並列レビュー」エントリがこれらの SSOT を参照するように更新してください。
In `@opencode/skills/harness-review/SKILL.md`:
- Around line 263-266: The fenced code block containing the lines "✅ 正しい: 4回の
MCP 呼び出しを1つのレスポンス内で並列実行" and "❌ 間違い: 1回の呼び出しで「全観点をレビューして」と依頼" is missing a
language tag (MD040); update the opening fence from ``` to ```text (or another
appropriate language) in SKILL.md so the block is fenced as ```text ... ``` and
keep the same closing fence, ensuring the code block renders with a language
specifier.
In `@skills/harness-review/SKILL.md`:
- Around line 263-266: The Markdown code fence shown (the triple backticks
surrounding the Japanese checklist block containing "✅ 正しい..." and "❌ 間違い...")
lacks a language tag and triggers markdownlint MD040; fix it by adding a
language identifier (e.g., ```text) immediately after the opening backticks so
the fenced block becomes ```text ... ```; update the fenced block in SKILL.md
where the checklist appears to include that language tag.
In `@skills/session-state/references/state-transition` 2.md:
- Around line 1-5: Rename the file currently named "state-transition 2.md" to
"state-transition.md" so it matches the expected SKILL reference path
"references/state-transition.md"; verify and update any references or imports
that point to "references/state-transition.md" and ensure the frontmatter name
("name: state-transition") remains correct after renaming so automatic loading
and links resolve properly.
In `@skills/session-state/SKILL` 2.md:
- Around line 1-19: The SKILL file is currently named "SKILL 2.md" which will
likely be skipped by the loader; rename skills/session-state/SKILL 2.md to
skills/session-state/SKILL.md (keep the name field "session-state") and ensure
the referenced document filename matches exactly: verify the link
references/state-transition.md points to an existing file and either rename that
file to state-transition.md or update the link in SKILL.md to the actual
filename so the reference and loader conventions align.
♻️ Duplicate comments (1)
skills/codex-review/references/codex-parallel-review.md (1)
17-29: フェンスコードに言語指定を付与してください（MD040）。既存指摘と同内容です。
Comment on lines
17
to
29
| ``` | ||
| Claude (オーケストレーター) | ||
| ↓ | ||
| 並列 MCP 呼び出し | ||
| ├── Security Expert | ||
| ├── Accessibility Expert | ||
| ├── Performance Expert | ||
| ├── Quality Expert | ||
| ├── SEO Expert | ||
| ├── Architect Expert | ||
| ├── Plan Reviewer Expert | ||
| └── Scope Analyst Expert | ||
| レビュータイプ判定 | ||
| ↓ | ||
| 並列 MCP 呼び出し (4 experts) | ||
| ├── Expert 1 | ||
| ├── Expert 2 | ||
| ├── Expert 3 | ||
| └── Expert 4 | ||
| ↓ | ||
| 結果統合 → コミット判定 | ||
| 結果統合 → 判定 | ||
| ``` |
There was a problem hiding this comment.
フェンスコードに言語指定を付与してください（MD040）。
🔧 修正案
-```
+```text
Claude (オーケストレーター)
↓
レビュータイプ判定
↓
並列 MCP 呼び出し (4 experts)
├── Expert 1
├── Expert 2
├── Expert 3
└── Expert 4
↓
結果統合 → 判定</details>
<details>
<summary>🤖 Prompt for AI Agents</summary>
In @opencode/skills/codex-review/references/codex-parallel-review.md around
lines 17 - 29, The fenced code block containing the ASCII flow diagram (starting
with the three backticks and the line "Claude (オーケストレーター)" and ending with three
backticks) is missing a language identifier which triggers MD040; fix it by
adding a language specifier (e.g., change the opening totext) so the
block becomes text ... and keep the diagram content unchanged to satisfy
the linter.
</details>
<!-- fingerprinting:phantom:triton:eagle -->
<!-- This is an auto-generated comment by CodeRabbit -->
| | **MCP セットアップ** | See [references/codex-mcp-setup.md](references/codex-mcp-setup.md) | | ||
| | **レビュー統合** | See [references/codex-review-integration.md](references/codex-review-integration.md) | | ||
| | **最大8並列レビュー** | See [references/codex-parallel-review.md](references/codex-parallel-review.md) | | ||
| | **4並列レビュー** | See [references/codex-parallel-review.md](references/codex-parallel-review.md) | |
There was a problem hiding this comment.
🧩 Analysis chain
🏁 Script executed:
#!/bin/bash
# Check if .claude/memory directory and files exist
echo "=== Checking .claude/memory structure ==="
find .claude/memory -type f 2>/dev/null | head -20
echo ""
echo "=== Searching for 4-parallel or parallel review references in SSOT files ==="
rg -n "4.*並列|4.*expert|並列レビュー" .claude/memory/ 2>/dev/null || echo "No matches found in .claude/memory"
echo ""
echo "=== Searching in decisions.md specifically ==="
if [ -f .claude/memory/decisions.md ]; then
cat .claude/memory/decisions.md
else
echo ".claude/memory/decisions.md not found"
fi
echo ""
echo "=== Searching in patterns.md specifically ==="
if [ -f .claude/memory/patterns.md ]; then
cat .claude/memory/patterns.md
else
echo ".claude/memory/patterns.md not found"
fi
echo ""
echo "=== Checking for referenced files ==="
ls -la opencode/skills/codex-review/references/codex-parallel-review.md 2>/dev/null || echo "references/codex-parallel-review.md not found"
echo ""
echo "=== Checking for accessibility-expert.md ==="
find opencode -name "*accessibility*" -type f 2>/dev/null | head -10Repository: Chachamaru127/claude-code-harness
Length of output: 739
🏁 Script executed:
#!/bin/bash
# Check .claude directory existence
echo "=== Checking .claude directory ==="
ls -la .claude 2>/dev/null || echo ".claude directory not found"
echo ""
echo "=== Reading SKILL.md around line 28 ==="
sed -n '20,40p' opencode/skills/codex-review/SKILL.md
echo ""
echo "=== Reading lines 49-53 for expert changes ==="
sed -n '45,60p' opencode/skills/codex-review/SKILL.md
echo ""
echo "=== Reading codex-parallel-review.md ==="
head -50 opencode/skills/codex-review/references/codex-parallel-review.mdRepository: Chachamaru127/claude-code-harness
Length of output: 2858
.claude/memory に SSOT ファイルが見つかりません。アーキテクチャドキュメントとして decisions.md と patterns.md を作成してください。
コーディングガイドラインに従い、opencode/**/*.md への変更は .claude/memory/decisions.md（Why）と .claude/memory/patterns.md（How）を SSOT として記載する必要があります。
以下の内容を記載してください:
- decisions.md: 「なぜ 4 並列レビューに変更したのか」（背景、意思決定根拠）
- patterns.md: 「4 並列レビューの実行パターン」（エキスパート定義、MCP 呼び出し手順、codex-parallel-review.md への参照）
🤖 Prompt for AI Agents
In `@opencode/skills/codex-review/SKILL.md` at line 28, opencode/**/*.md の変更に対して
SSOT が存在しないため、.claude/memory/decisions.md（Why）と
.claude/memory/patterns.md（How）を作成してください; decisions.md に「なぜ 4
並列レビューに変更したのか」という背景と意思決定根拠（利点・トレードオフ・関係者の合意）を明確に記載し、patterns.md に「4
並列レビューの実行パターン」（エキスパートの役割定義、MCP 呼び出し手順、参照として references/codex-parallel-review.md
へのリンクと実践フロー）を具体的な手順として記載してください; 作成後、opencode/skills/codex-review/SKILL.md
の該当箇所にある「4並列レビュー」エントリがこれらの SSOT を参照するように更新してください。
Comment on lines
263
to
266
| ``` | ||
| ✅ 正しい: 8回の MCP 呼び出しを1つのレスポンス内で並列実行 | ||
| ✅ 正しい: 4回の MCP 呼び出しを1つのレスポンス内で並列実行 | ||
| ❌ 間違い: 1回の呼び出しで「全観点をレビューして」と依頼 | ||
| ``` |
There was a problem hiding this comment.
フェンスコードに言語指定を付与してください（MD040）。
🔧 修正案
-```
+```text
✅ 正しい: 4回の MCP 呼び出しを1つのレスポンス内で並列実行
❌ 間違い: 1回の呼び出しで「全観点をレビューして」と依頼</details>
<!-- suggestion_start -->
<details>
<summary>📝 Committable suggestion</summary>
> ‼️ **IMPORTANT**
> Carefully review the code before committing. Ensure that it accurately replaces the highlighted code, contains no missing lines, and has no issues with indentation. Thoroughly test & benchmark the code to ensure it meets the requirements.
```suggestion
🧰 Tools
🪛 markdownlint-cli2 (0.18.1)
263-263: Fenced code blocks should have a language specified
(MD040, fenced-code-language)
🤖 Prompt for AI Agents
In `@opencode/skills/harness-review/SKILL.md` around lines 263 - 266, The fenced
code block containing the lines "✅ 正しい: 4回の MCP 呼び出しを1つのレスポンス内で並列実行" and "❌ 間違い:
1回の呼び出しで「全観点をレビューして」と依頼" is missing a language tag (MD040); update the opening
fence from ``` to ```text (or another appropriate language) in SKILL.md so the
block is fenced as ```text ... ``` and keep the same closing fence, ensuring the
code block renders with a language specifier.
Comment on lines
263
to
266
| ``` | ||
| ✅ 正しい: 8回の MCP 呼び出しを1つのレスポンス内で並列実行 | ||
| ✅ 正しい: 4回の MCP 呼び出しを1つのレスポンス内で並列実行 | ||
| ❌ 間違い: 1回の呼び出しで「全観点をレビューして」と依頼 | ||
| ``` |
There was a problem hiding this comment.
コードフェンスに言語指定を追加してください。
markdownlint の MD040 対応が必要です。
🔧 修正案
-```
+```text📝 Committable suggestion
‼️ IMPORTANT
Carefully review the code before committing. Ensure that it accurately replaces the highlighted code, contains no missing lines, and has no issues with indentation. Thoroughly test & benchmark the code to ensure it meets the requirements.
Suggested change
| ``` | |
| ✅ 正しい: 8回の MCP 呼び出しを1つのレスポンス内で並列実行 | |
| ✅ 正しい: 4回の MCP 呼び出しを1つのレスポンス内で並列実行 | |
| ❌ 間違い: 1回の呼び出しで「全観点をレビューして」と依頼 | |
| ``` |
🧰 Tools
🪛 markdownlint-cli2 (0.18.1)
263-263: Fenced code blocks should have a language specified
(MD040, fenced-code-language)
🤖 Prompt for AI Agents
In `@skills/harness-review/SKILL.md` around lines 263 - 266, The Markdown code
fence shown (the triple backticks surrounding the Japanese checklist block
containing "✅ 正しい..." and "❌ 間違い...") lacks a language tag and triggers
markdownlint MD040; fix it by adding a language identifier (e.g., ```text)
immediately after the opening backticks so the fenced block becomes ```text ...
```; update the fenced block in SKILL.md where the checklist appears to include
that language tag.
Comment on lines
+1
to
+5
| --- | ||
| name: state-transition | ||
| description: "Execute session state transitions using session-state.sh" | ||
| allowed-tools: [Read, Bash] | ||
| --- |
There was a problem hiding this comment.
参照パスとファイル名が一致していません。
SKILL からの参照想定が references/state-transition.md のため、state-transition 2.md のままだとリンク切れや自動読込失敗の可能性があります。state-transition.md へリネームして統一してください。
🤖 Prompt for AI Agents
In `@skills/session-state/references/state-transition` 2.md around lines 1 - 5,
Rename the file currently named "state-transition 2.md" to "state-transition.md"
so it matches the expected SKILL reference path
"references/state-transition.md"; verify and update any references or imports
that point to "references/state-transition.md" and ensure the frontmatter name
("name: state-transition") remains correct after renaming so automatic loading
and links resolve properly.
Comment on lines
+1
to
+19
| --- | ||
| name: session-state | ||
| version: 1.0.0 | ||
| description: "Manages session state transitions per SESSION_ORCHESTRATION.md. Internal use only." | ||
| allowed-tools: [Read, Bash] | ||
| dependencies: [] | ||
| --- | ||
|
|
||
| # Session State Skill | ||
|
|
||
| セッション状態の遷移を管理する内部スキル。 | ||
| `docs/SESSION_ORCHESTRATION.md` に定義された状態機械に従って遷移を検証・実行する。 | ||
|
|
||
| ## 機能詳細 | ||
|
|
||
| | 機能 | 詳細 | | ||
| |------|------| | ||
| | **状態遷移** | See [references/state-transition.md](references/state-transition.md) | | ||
|
|
There was a problem hiding this comment.
SKILL ファイル名が検出対象外になる懸念
一般的にスキル読込は skills/*/SKILL.md を前提にしているため、SKILL 2.md のままだと読み込まれない可能性があります。あわせて参照先が references/state-transition.md なので、参照ファイル名も揃える形でリネームしてください。
🤖 Prompt for AI Agents
In `@skills/session-state/SKILL` 2.md around lines 1 - 19, The SKILL file is
currently named "SKILL 2.md" which will likely be skipped by the loader; rename
skills/session-state/SKILL 2.md to skills/session-state/SKILL.md (keep the name
field "session-state") and ensure the referenced document filename matches
exactly: verify the link references/state-transition.md points to an existing
file and either rename that file to state-transition.md or update the link in
SKILL.md to the actual filename so the reference and loader conventions align.
- work.md: review skill → harness-review, 8-parallel → 4-parallel - commit-judgment-logic.md: update to 3 review types with 4 experts each - execute-commit.md: Reviewed-by: Codex (8 experts) → (4 experts) - codex-review-integration.md: review スキル → harness-review スキル - harness-review SKILL.md: fix experts/ path to ../codex-review/references/experts/ - codex-parallel-review.md: add Plan/Scope output format constraints Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
5 tasks done
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary by CodeRabbit
新機能
ドキュメント
雑務
✏️ Tip: You can customize this high-level summary in your review settings.