Argent MCP ツールキットのスキル・エージェント・設定ファイルを追加#6420
Merged
Merged
Conversation
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Contributor
📝 WalkthroughWalkthroughArgent MCPの接続設定と運用規約を追加し、iOS・Android・Chromium・TV操作、React Native開発、Metroデバッグ、UIテスト、フロー再生、プロファイリング、最適化のスキル文書を整備しました。 ChangesArgent運用基盤
Estimated code review effort: 4 (Complex) | ~60 minutes Poem
🚥 Pre-merge checks | ✅ 5✅ Passed checks (5 passed)
✨ Finishing Touches📝 Generate docstrings
🧪 Generate unit tests (beta)
Comment |
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
Contributor
There was a problem hiding this comment.
Actionable comments posted: 14
Note
Due to the large number of review comments, Critical, Major severity comments were prioritized as inline comments.
Caution
Some comments are outside the diff and can’t be posted inline due to platform limitations.
⚠️ Outside diff range comments (1)
.agents/skills/argent-screenshot-diff/SKILL.md (1)
81-90: 📐 Maintainability & Code Quality | 🟡 Minor | ⚡ Quick winコードフェンスに言語タグを追加してください。
この例はMarkdownコードフェンスに言語指定がなく、
markdownlint-cli2のMD040違反になります。実行可能なJSONでないため、textなど適切なタグを付けてください。🤖 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 @.agents/skills/argent-screenshot-diff/SKILL.md around lines 81 - 90, Update the Markdown code fence in the example within the skill documentation to include an appropriate language tag, such as text, so it no longer violates markdownlint MD040 while preserving the example’s non-executable JSON content.Sources: Coding guidelines, Linters/SAST tools
🟡 Minor comments (15)
.agents/skills/argent-create-flow/SKILL.md-203-206 (1)
203-206: 🎯 Functional Correctness | 🟡 Minor | ⚡ Quick win修正戦略の判定境界が重複しています。
3 ステップの不具合が Strategy 3 と Strategy 4 の両方に該当します。Strategy 4 を
4+ steps brokenに変更するなど、各入力が一意に分類されるようにしてください。🤖 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 @.agents/skills/argent-create-flow/SKILL.md around lines 203 - 206, 修正戦略の判定基準で「3 steps broken」がStrategy 3とStrategy 4に重複するため、Strategy 4の条件を「4+ steps broken」に更新し、各ステップ数が一意にStrategy 3またはStrategy 4へ分類されるようにしてください。.claude/skills/argent-metro-debugger/SKILL.md-99-99 (1)
99-99: 📐 Maintainability & Code Quality | 🟡 Minor | ⚡ Quick winログレベル例の空白を Markdownlint 対応にしてください。
WARNとLOGはコードスパン内の末尾空白として MD038 に検出されます。WARN␠/LOG␠␠のように空白を可視化してください。As per coding guidelines, markdownlint-cli2 の規約に従い、コードスパン内の不要な空白を避けます。
🤖 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 @.claude/skills/argent-metro-debugger/SKILL.md at line 99, Update the log-level example in the table near the level-format documentation to make the padding spaces in WARN and LOG visible rather than trailing inside code spans, while preserving their uppercase five-character format.Sources: Coding guidelines, Linters/SAST tools
.agents/skills/argent-create-flow/SKILL.md-46-63 (1)
46-63: 📐 Maintainability & Code Quality | 🟡 Minor | ⚡ Quick winすべての fenced code block に言語タグを付けてください。
Markdownlint-cli2 の MD040 により、6 個のコードフェンスが lint 違反になります。ツール引数の例は
yaml、セッション実行例はtextまたはconsoleなどを指定してください。As per coding guidelines, Markdown の fenced code block には言語タグを付けます。
Also applies to: 82-102
🤖 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 @.agents/skills/argent-create-flow/SKILL.md around lines 46 - 63, Update every fenced code block in SKILL.md, including the additional blocks around the session execution examples, to include an appropriate language tag. Use yaml for tool argument examples and text or console for session execution examples, ensuring no unlabeled fences remain.Sources: Coding guidelines, Linters/SAST tools
.claude/skills/argent-metro-debugger/SKILL.md-82-89 (1)
82-89: 🚀 Performance & Scalability | 🟡 Minor | ⚡ Quick winログ読み出し手順の
Readと「直接読まない」を統一してください。Line 87 と Quick Reference は
Grep/Readを案内していますが、Line 109 はログファイルを直接Readしないよう要求しています。全量読み込みを禁止するなら、全箇所を「制限付きGrepまたは shell コマンド」に統一し、許可する場合はReadの行数制限を明記してください。Also applies to: 107-110, 128-128
🤖 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 @.claude/skills/argent-metro-debugger/SKILL.md around lines 82 - 89, 統一して、`debugger-log-registry` の結果で得たログファイルは全量 `Read` せず、制限付きの `Grep` または範囲を指定した shell コマンドで検索する手順に更新してください。`Workflow`、Quick Reference、および該当するログ読み出し指示から無制限の `Read` 案内を削除し、大規模ログの `Explore` 委譲手順も同じ制約に合わせてください。.agents/skills/argent-metro-debugger/references/source-maps.md-15-24 (1)
15-24: 📐 Maintainability & Code Quality | 🟡 Minor | ⚡ Quick win埋め込み JavaScript の引用符をシングルクォートに統一してください。
- presets: ["babel-preset-expo"], // or 'module:`@react-native/babel-preset`' + presets: ['babel-preset-expo'], // or 'module:`@react-native/babel-preset`' plugins: [ - "`@babel/plugin-transform-react-jsx-source`", // enables _debugSource on fibers + '`@babel/plugin-transform-react-jsx-source`', // enables _debugSource on fibersAs per coding guidelines, JavaScript の文字列にはシングルクォートを使用します。
🤖 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 @.agents/skills/argent-metro-debugger/references/source-maps.md around lines 15 - 24, Update the embedded JavaScript configuration in the Babel module export to use single quotes for all string literals, including the preset and plugin entries, while leaving the configuration behavior unchanged.Source: Coding guidelines
.agents/skills/argent-react-native-optimization/references/fix-reference.md-7-8 (1)
7-8: 🚀 Performance & Scalability | 🟡 Minor | ⚡ Quick winReact Compiler の扱いと
useCallbackの説明を修正してください。
.agents/skills/argent-react-native-optimization/references/fix-reference.md:7-8のSkip if React Compiler activeは一律スキップに見えるため、compiler がそのコンポーネントを処理している場合のみ省略し、bailout がある場合は手動 memoization を案内する表現にしてください。useCallbackも子のReact.memoに限らず、Hook の依存配列安定化にも使えます。🤖 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 @.agents/skills/argent-react-native-optimization/references/fix-reference.md around lines 7 - 8, 修正対象は最適化判断表の React Compiler と useCallback の説明です。React Compiler が対象コンポーネントを実際に処理している場合のみ React.memo を省略し、bailout 時は手動 memoization を案内する表現に更新してください。また、useCallback の用途を React.memo 化された子への関数安定化に限定せず、Hook の依存配列を安定化する用途も含めて説明してください。.claude/skills/argent-create-flow/SKILL.md-46-49 (1)
46-49: 📐 Maintainability & Code Quality | 🟡 Minor | ⚡ Quick winコードフェンスに言語タグを付けてください。
MarkdownlintのMD040違反です。シェル例には
shell、YAML例にはyaml、JSON例にはjsonなど、各フェンスに適切な言語タグを追加してください。Also applies to: 51-54, 56-59, 61-64, 82-91, 95-102
🤖 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 @.claude/skills/argent-create-flow/SKILL.md around lines 46 - 49, Update every fenced code block identified in the skill document, including the shown launch-app example and the additional affected ranges, to include an appropriate language tag such as yaml, json, or shell. Use the tag matching each block’s actual content and preserve the examples unchanged.Sources: Coding guidelines, Linters/SAST tools
.claude/agents/argent-environment-inspector.md-20-20 (1)
20-20: 📐 Maintainability & Code Quality | 🟡 Minor | ⚡ Quick win重複した句読点を修正してください。
**Your first task ... is.**.はピリオドが重複しています。🤖 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 @.claude/agents/argent-environment-inspector.md at line 20, 修正対象の文「Your first task is to determine what kind of project this is.」末尾にある重複したピリオドを1つ削除し、文末の句読点を1つだけ残してください。.claude/skills/argent-lens/SKILL.md-81-101 (1)
81-101: 📐 Maintainability & Code Quality | 🟡 Minor | ⚡ Quick winコードフェンスに言語タグを追加してください。
この例は JSON ではなくツール呼び出しの擬似コードなので、
```textなどの言語タグを付けて Markdown lint に準拠させてください。As per coding guidelines: Markdown の fenced code block には言語タグを付けること。
🤖 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 @.claude/skills/argent-lens/SKILL.md around lines 81 - 101, SKILL.md の該当するツール呼び出し擬似コードのコードフェンスに、内容に適した言語タグとして text を追加してください。Sources: Coding guidelines, Linters/SAST tools
.claude/skills/argent-device-interact/SKILL.md-131-137 (1)
131-137: 🎯 Functional Correctness | 🟡 Minor | ⚡ Quick win
gesture-pinchの正規化説明から非座標パラメータを除外してください。「All values are normalized」とありますが、同じ節で
angle: 90とdurationMs: 500を例示しています。centerX、centerY、startDistance、endDistanceなどの座標・距離だけが正規化対象だと明記してください。🤖 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 @.claude/skills/argent-device-interact/SKILL.md around lines 131 - 137, Update the gesture-pinch documentation to state that normalization applies only to coordinate and distance parameters such as centerX, centerY, startDistance, and endDistance; exclude angle and durationMs from the “All values are normalized” wording while preserving their existing examples and defaults..claude/skills/argent-react-native-optimization/references/fix-reference.md-7-11 (1)
7-11: 🎯 Functional Correctness | 🟡 Minor | ⚡ Quick win説明を条件付きに修正してください
8行目のuseCallbackはReact.memoの子への props 安定化だけでなく、useEffectなどの依存値安定化にも使えます。11行目のuseNativeDriverもtransform/opacityに限定せず、非レイアウト系プロパティに絞る表現が正確です。🤖 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 @.claude/skills/argent-react-native-optimization/references/fix-reference.md around lines 7 - 11, Update the guidance in the rows describing unstable callbacks and JS-thread animation jank: state that useCallback can stabilize props for React.memo children or dependency values such as useEffect, and describe useNativeDriver as supporting non-layout properties rather than only transform and opacity..agents/skills/argent-react-native-optimization/SKILL.md-26-32 (1)
26-32: 📐 Maintainability & Code Quality | 🟡 Minor | ⚡ Quick winコードフェンスに言語タグを追加してください。
チェックリスト用のフェンスに言語指定がなく、MD040違反になります。
markdownまたはtextを指定してください。🤖 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 @.agents/skills/argent-react-native-optimization/SKILL.md around lines 26 - 32, Update the code fence surrounding the “Optimization Progress” checklist in the skill documentation to include an explicit markdown or text language tag, resolving the MD040 violation while preserving the checklist content.Sources: Coding guidelines, Linters/SAST tools
.claude/skills/argent-native-profiler/SKILL.md-79-81 (1)
79-81: 🎯 Functional Correctness | 🟡 Minor | ⚡ Quick winAndroidの例で
<UDID>を固定しないでください。この手順はAndroidにも適用されますが、Androidの
device_idはserialです。device_id=<device_id>に変更するか、iOSはUDID、Androidはserialと明記してください。🤖 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 @.claude/skills/argent-native-profiler/SKILL.md around lines 79 - 81, Update the profiler-load instructions in the Android-applicable workflow to use device_id=<device_id> instead of the fixed <UDID> placeholder, and explicitly state that iOS uses the UDID while Android uses the device serial..agents/skills/argent-lens/SKILL.md-70-100 (1)
70-100: 🎯 Functional Correctness | 🟡 Minor | ⚡ Quick winサンプルにも各要素2バリアント以上を追加してください。
Primary CTAは1バリアントしかなく、70行目のルールに違反しています。サンプルに別バリアントを追加するか、ルールを「選択対象の要素のみ2件以上」と明確化してください。🤖 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 @.agents/skills/argent-lens/SKILL.md around lines 70 - 100, Update the “Primary CTA” example in the SKILL.md sample to include at least two distinct variants, each with its own built, visibly different previewImage, while preserving the existing propose_variant and selection flow. Alternatively, explicitly revise the stated rule to limit the two-variant requirement to selectable elements, but keep the sample consistent with that clarified rule..claude/skills/argent-metro-debugger/references/source-maps.md-19-21 (1)
19-21: 📐 Maintainability & Code Quality | 🟡 Minor | ⚡ Quick winJavaScript 例の引用符をシングルクォートに統一してください。
リポジトリのコーディング規約に反して、Babel 設定例でダブルクォートを使用しています。
修正例
- presets: ["babel-preset-expo"], // or 'module:`@react-native/babel-preset`' + presets: ['babel-preset-expo'], // or 'module:`@react-native/babel-preset`' plugins: [ - "`@babel/plugin-transform-react-jsx-source`", // enables _debugSource on fibers + '`@babel/plugin-transform-react-jsx-source`', // enables _debugSource on fibers🤖 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 @.claude/skills/argent-metro-debugger/references/source-maps.md around lines 19 - 21, Update the Babel configuration example in the presets and plugins entries to use single quotes consistently instead of double quotes, while preserving the existing values and comments.Source: Coding guidelines
🧹 Nitpick comments (4)
.claude/skills/argent-react-native-optimization/SKILL.md (1)
26-32: 📐 Maintainability & Code Quality | 🔵 Trivial | ⚡ Quick winフェンスコードブロックに言語タグがない(MD040)
26行目のチェックリストのコードフェンスに言語指定がありません。ガイドラインの「fenced code blocks should have a language specified」に反します。
textタグを付与してください。🤖 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 @.claude/skills/argent-react-native-optimization/SKILL.md around lines 26 - 32, Update the fenced code block containing the “Optimization Progress” checklist in SKILL.md to specify the text language tag, while leaving the checklist content unchanged.Sources: Coding guidelines, Linters/SAST tools
.codex/config.toml (2)
25-27: 🔒 Security & Privacy | 🔵 Trivial全ツール一律
"approve"は読み取り専用ツールにも過剰な確認プロンプトを強制する
screenshot(25行目付近)やdescribeのような読み取り専用ツールと、debugger-evaluate(82行目付近、任意のJS評価)やstop-all-simulator-servers(187行目付近)のような影響の大きい操作が同じ"approve"で扱われています。SKILL.md群は1フローで数十回のツール呼び出しを前提にしており、毎回の確認が運用上の摩擦になりえます。安全な読み取り系は"auto"にする、またはサーバー単位でdefault_tools_approval_mode = "writes"を使い書き込み系のみ確認を求める設計も検討してください。Also applies to: 82-84, 187-188
🤖 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 @.codex/config.toml around lines 25 - 27, Update the MCP approval configuration for the read-only screenshot and describe tools to use "auto" approval, while retaining confirmation for high-impact tools such as debugger-evaluate and stop-all-simulator-servers. Apply the same distinction consistently across the referenced tool entries rather than using a blanket "approve" setting.
6-216: 📐 Maintainability & Code Quality | 🔵 Trivial | ⚡ Quick win55個の個別
approval_mode設定はサーバー単位のdefault_tools_approval_modeに置き換え可能すべてのツールで値が
"approve"と同一のため、[mcp_servers.argent]直下にdefault_tools_approval_mode = "approve"を1行追加すれば同じ挙動を実現できます。約210行を1行に削減でき、将来argentに新規ツールが追加された際に個別設定を追加し忘れて意図せず既定挙動に戻ってしまうリスクも防げます。♻️ 提案する修正
[mcp_servers.argent] command = "argent" args = [ "mcp" ] - -[mcp_servers.argent.tools.list-devices] -approval_mode = "approve" - -[mcp_servers.argent.tools.boot-device] -approval_mode = "approve" - -... (以下、全55ブロックを削除) ... +default_tools_approval_mode = "approve"🤖 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 @.codex/config.toml around lines 6 - 216, Replace the repeated approval_mode entries under mcp_servers.argent.tools with a single default_tools_approval_mode = "approve" setting directly under mcp_servers.argent, removing the individual tool-level configurations while preserving approval behavior for all current and future tools..claude/skills/argent-test-ui-flow/SKILL.md (1)
42-99: 📐 Maintainability & Code Quality | 🔵 Trivial | ⚡ Quick winフェンスコードブロックに言語タグがない(MD040)
42, 58, 70, 81, 94行目のコードフェンスに言語指定がありません。コーディングガイドラインでは「fenced code blocks should have a language specified」ことが求められています。
textなどのタグを付与してください。🤖 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 @.claude/skills/argent-test-ui-flow/SKILL.md around lines 42 - 99, コードフェンスで囲まれた各例に言語タグがないため、SKILL.md内の該当するすべてのフェンス（Goal/Steps、Login flow、Scroll and navigation、Visual behavior check、Wait for a loading spinner）に内容に適した`text`などの言語指定を追加してください。Sources: Coding guidelines, Linters/SAST tools
🤖 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 @.agents/skills/argent-create-flow/SKILL.md:
- Line 25: Update the flow-execute example in the skill instructions to remove
project_root, leaving only name and, when needed, prerequisiteAcknowledged. Keep
project_root specified exclusively for flow-start-recording, consistent with the
surrounding documentation.
In @.agents/skills/argent-lens/SKILL.md:
- Around line 58-62: Clarify the await_user_selection guidance to state it is
called once per completed selection round, but must be called again whenever it
returns status "pending" until a non-pending result is received. Preserve the
existing handling for "completed" and "no_proposals".
In @.agents/skills/argent-metro-debugger/SKILL.md:
- Around line 82-110:
ログ読取手順をgrepなどの限定検索に統一し、直接Readを禁止する方針と矛盾しないよう更新してください。.agents/skills/argent-metro-debugger/SKILL.md（82-110行）では「GrepまたはRead」およびRead禁止の記述を一貫した限定読取方針に整理し、.agents/skills/argent-react-native-app-workflow/SKILL.md（162-175行）と.claude/skills/argent-react-native-app-workflow/SKILL.md（162-175行）も同じ手順・制約に合わせてください。
In @.agents/skills/argent-native-profiler/SKILL.md:
- Around line 54-61: プロファイルデータの保存・再利用を明示的な opt-in
に統一し、プロジェクト分離、アクセス権、保持期限、削除手順を共通ポリシーとして記載してください。.agents/skills/argent-native-profiler/SKILL.md
54-61 の native-profiler-start/stop 手順に保存ポリシーと cleanup
を追加し、.claude/skills/argent-react-native-profiler/references/diagnostic-tools.md
88-94
では保存セッションの一覧・再読込を無条件に案内しないよう変更してください。.agents/skills/argent-react-native-profiler/SKILL.md
129-138 の React セッション再読込にも同じ同意・保持ルールを適用してください。
In @.agents/skills/argent-react-native-app-workflow/SKILL.md:
- Line 114: 通常の復旧手順ではロックファイルを削除せず、両方のサイトで package ロックファイルと Podfile.lock
を保持するよう更新してください。.agents/skills/argent-react-native-app-workflow/SKILL.md
の114行目では npm ci または immutable install
を優先し、.claude/skills/argent-react-native-app-workflow/SKILL.md
の114行目にも同じ変更を適用してください。
- Line 55:
Metroポートを8081に固定せず、設定から検出したMetroポートを使う手順に更新してください。.agents/skills/argent-react-native-app-workflow/SKILL.mdの55行目と62行目、および.claude/skills/argent-react-native-app-workflow/SKILL.mdの55行目と62行目で、adb
reverseとチェックリストが同じ設定済みポート変数を参照するように変更してください。
In @.agents/skills/argent-react-native-optimization/references/lint-rules.md:
- Around line 3-4:
標準の検証ツール契約をBiomeへ統一する。.agents/skills/argent-react-native-optimization/references/lint-rules.mdの3-4行ではESLintプラグイン導入指示を削除し、Biomeの導入・設定確認に置き換える。同ファイル49-55行ではESLint実行手順をBiomeのlintおよびformat手順へ変更する。.claude/agents/references/quality-control-checklist.mdの3-12行では無効なbool値を有効なJSON値に修正し、eslint/prettier項目をBiomeの出力契約へ置き換える。
In @.agents/skills/argent-react-native-optimization/SKILL.md:
- Around line 34-42: Update the “Phase 1: Lint sweep” instructions to use the
repository’s Biome-based lint script or Biome command instead of ESLint.
Preserve the existing project-root execution and one-sub-agent-per-file
workflow, and update the referenced lint procedure if it also specifies ESLint.
In @.claude/agents/argent-environment-inspector.md:
- Around line 42-48: Update the project classification rules so Expo detection
takes precedence over React Native detection, checking for expo in dependencies
or an app.json expo key first. Alternatively, explicitly exclude Expo projects
from the react-native condition while preserving the existing native iOS, native
Android, Flutter, and fallback classifications.
In @.claude/rules/argent.md:
- Around line 71-83: Update the general_rules simulator/emulator command
guidance to resolve the contradiction between the prohibition on direct
interactions and the fallback allowing xcrun or adb. Explicitly permit only
necessary non-interactive device/setup commands, including adb reverse and adb
emu kill, while continuing to require Argent MCP tools for direct UI
interactions and gestures.
In @.claude/skills/argent-create-flow/SKILL.md:
- Around line 82-90: Update the open-settings example in the flow recording
documentation to add a UI exploration step, such as describe, immediately before
each gesture-tap step. Ensure both the “Tap General” and “Tap About” actions use
fresh UI exploration before tapping, rather than relying only on guessed
coordinates.
In @.claude/skills/argent-lens/SKILL.md:
- Around line 56-62: Update the “Step 3 — Await the human's decision”
instructions to clarify that await_user_selection is called at the end of each
round, and retried within the same round while status is "pending". Keep the
existing "completed" and "no_proposals" handling unchanged, and remove the
contradictory implication that it is called only once overall.
In @.claude/skills/argent-react-native-optimization/references/lint-rules.md:
- Line 4: Update the fallback ESLint installation instructions to include every
dependency referenced by the listed rules, including eslint-plugin-react,
eslint-plugin-react-native, eslint-plugin-react-hooks, eslint-plugin-react-perf,
and the TypeScript parser when required; alternatively remove the fallback
procedure.
- Around line 3-4: Update the lint procedure in the referenced instructions to
use the repository-standard Biome formatter and linter commands from the project
root, removing the temporary ESLint configuration, npx eslint invocation, and
eslint-plugin-react-perf installation guidance. If ESLint remains, explicitly
limit it to auxiliary use and document why it is needed.
---
Outside diff comments:
In @.agents/skills/argent-screenshot-diff/SKILL.md:
- Around line 81-90: Update the Markdown code fence in the example within the
skill documentation to include an appropriate language tag, such as text, so it
no longer violates markdownlint MD040 while preserving the example’s
non-executable JSON content.
---
Minor comments:
In @.agents/skills/argent-create-flow/SKILL.md:
- Around line 203-206: 修正戦略の判定基準で「3 steps broken」がStrategy 3とStrategy
4に重複するため、Strategy 4の条件を「4+ steps broken」に更新し、各ステップ数が一意にStrategy 3またはStrategy
4へ分類されるようにしてください。
- Around line 46-63: Update every fenced code block in SKILL.md, including the
additional blocks around the session execution examples, to include an
appropriate language tag. Use yaml for tool argument examples and text or
console for session execution examples, ensuring no unlabeled fences remain.
In @.agents/skills/argent-lens/SKILL.md:
- Around line 70-100: Update the “Primary CTA” example in the SKILL.md sample to
include at least two distinct variants, each with its own built, visibly
different previewImage, while preserving the existing propose_variant and
selection flow. Alternatively, explicitly revise the stated rule to limit the
two-variant requirement to selectable elements, but keep the sample consistent
with that clarified rule.
In @.agents/skills/argent-metro-debugger/references/source-maps.md:
- Around line 15-24: Update the embedded JavaScript configuration in the Babel
module export to use single quotes for all string literals, including the preset
and plugin entries, while leaving the configuration behavior unchanged.
In @.agents/skills/argent-react-native-optimization/references/fix-reference.md:
- Around line 7-8: 修正対象は最適化判断表の React Compiler と useCallback の説明です。React
Compiler が対象コンポーネントを実際に処理している場合のみ React.memo を省略し、bailout 時は手動 memoization
を案内する表現に更新してください。また、useCallback の用途を React.memo 化された子への関数安定化に限定せず、Hook
の依存配列を安定化する用途も含めて説明してください。
In @.agents/skills/argent-react-native-optimization/SKILL.md:
- Around line 26-32: Update the code fence surrounding the “Optimization
Progress” checklist in the skill documentation to include an explicit markdown
or text language tag, resolving the MD040 violation while preserving the
checklist content.
In @.claude/agents/argent-environment-inspector.md:
- Line 20: 修正対象の文「Your first task is to determine what kind of project this
is.」末尾にある重複したピリオドを1つ削除し、文末の句読点を1つだけ残してください。
In @.claude/skills/argent-create-flow/SKILL.md:
- Around line 46-49: Update every fenced code block identified in the skill
document, including the shown launch-app example and the additional affected
ranges, to include an appropriate language tag such as yaml, json, or shell. Use
the tag matching each block’s actual content and preserve the examples
unchanged.
In @.claude/skills/argent-device-interact/SKILL.md:
- Around line 131-137: Update the gesture-pinch documentation to state that
normalization applies only to coordinate and distance parameters such as
centerX, centerY, startDistance, and endDistance; exclude angle and durationMs
from the “All values are normalized” wording while preserving their existing
examples and defaults.
In @.claude/skills/argent-lens/SKILL.md:
- Around line 81-101: SKILL.md の該当するツール呼び出し擬似コードのコードフェンスに、内容に適した言語タグとして text
を追加してください。
In @.claude/skills/argent-metro-debugger/references/source-maps.md:
- Around line 19-21: Update the Babel configuration example in the presets and
plugins entries to use single quotes consistently instead of double quotes,
while preserving the existing values and comments.
In @.claude/skills/argent-metro-debugger/SKILL.md:
- Line 99: Update the log-level example in the table near the level-format
documentation to make the padding spaces in WARN and LOG visible rather than
trailing inside code spans, while preserving their uppercase five-character
format.
- Around line 82-89: 統一して、`debugger-log-registry` の結果で得たログファイルは全量 `Read`
せず、制限付きの `Grep` または範囲を指定した shell コマンドで検索する手順に更新してください。`Workflow`、Quick
Reference、および該当するログ読み出し指示から無制限の `Read` 案内を削除し、大規模ログの `Explore`
委譲手順も同じ制約に合わせてください。
In @.claude/skills/argent-native-profiler/SKILL.md:
- Around line 79-81: Update the profiler-load instructions in the
Android-applicable workflow to use device_id=<device_id> instead of the fixed
<UDID> placeholder, and explicitly state that iOS uses the UDID while Android
uses the device serial.
In @.claude/skills/argent-react-native-optimization/references/fix-reference.md:
- Around line 7-11: Update the guidance in the rows describing unstable
callbacks and JS-thread animation jank: state that useCallback can stabilize
props for React.memo children or dependency values such as useEffect, and
describe useNativeDriver as supporting non-layout properties rather than only
transform and opacity.
---
Nitpick comments:
In @.claude/skills/argent-react-native-optimization/SKILL.md:
- Around line 26-32: Update the fenced code block containing the “Optimization
Progress” checklist in SKILL.md to specify the text language tag, while leaving
the checklist content unchanged.
In @.claude/skills/argent-test-ui-flow/SKILL.md:
- Around line 42-99:
コードフェンスで囲まれた各例に言語タグがないため、SKILL.md内の該当するすべてのフェンス（Goal/Steps、Login flow、Scroll and
navigation、Visual behavior check、Wait for a loading
spinner）に内容に適した`text`などの言語指定を追加してください。
In @.codex/config.toml:
- Around line 25-27: Update the MCP approval configuration for the read-only
screenshot and describe tools to use "auto" approval, while retaining
confirmation for high-impact tools such as debugger-evaluate and
stop-all-simulator-servers. Apply the same distinction consistently across the
referenced tool entries rather than using a blanket "approve" setting.
- Around line 6-216: Replace the repeated approval_mode entries under
mcp_servers.argent.tools with a single default_tools_approval_mode = "approve"
setting directly under mcp_servers.argent, removing the individual tool-level
configurations while preserving approval behavior for all current and future
tools.
🪄 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: 6f840dd5-e947-44a0-a389-08432bcf5b87
📒 Files selected for processing (48)
.agents/skills/argent-android-emulator-setup/SKILL.md.agents/skills/argent-create-flow/SKILL.md.agents/skills/argent-device-interact/SKILL.md.agents/skills/argent-device-interact/references/gesture-examples.md.agents/skills/argent-ios-simulator-setup/SKILL.md.agents/skills/argent-lens/SKILL.md.agents/skills/argent-metro-debugger/SKILL.md.agents/skills/argent-metro-debugger/references/failure-scenarios.md.agents/skills/argent-metro-debugger/references/source-maps.md.agents/skills/argent-native-profiler/SKILL.md.agents/skills/argent-react-native-app-workflow/SKILL.md.agents/skills/argent-react-native-optimization/SKILL.md.agents/skills/argent-react-native-optimization/references/fix-reference.md.agents/skills/argent-react-native-optimization/references/lint-rules.md.agents/skills/argent-react-native-optimization/references/semantic-checklist.md.agents/skills/argent-react-native-profiler/SKILL.md.agents/skills/argent-react-native-profiler/references/diagnostic-tools.md.agents/skills/argent-screenshot-diff/SKILL.md.agents/skills/argent-test-ui-flow/SKILL.md.agents/skills/argent-tv-interact/SKILL.md.claude/agents/argent-environment-inspector.md.claude/agents/references/quality-control-checklist.md.claude/rules/argent.md.claude/settings.json.claude/skills/argent-android-emulator-setup/SKILL.md.claude/skills/argent-create-flow/SKILL.md.claude/skills/argent-device-interact/SKILL.md.claude/skills/argent-device-interact/references/gesture-examples.md.claude/skills/argent-ios-simulator-setup/SKILL.md.claude/skills/argent-lens/SKILL.md.claude/skills/argent-metro-debugger/SKILL.md.claude/skills/argent-metro-debugger/references/failure-scenarios.md.claude/skills/argent-metro-debugger/references/source-maps.md.claude/skills/argent-native-profiler/SKILL.md.claude/skills/argent-react-native-app-workflow/SKILL.md.claude/skills/argent-react-native-optimization/SKILL.md.claude/skills/argent-react-native-optimization/references/fix-reference.md.claude/skills/argent-react-native-optimization/references/lint-rules.md.claude/skills/argent-react-native-optimization/references/semantic-checklist.md.claude/skills/argent-react-native-profiler/SKILL.md.claude/skills/argent-react-native-profiler/references/diagnostic-tools.md.claude/skills/argent-screenshot-diff/SKILL.md.claude/skills/argent-test-ui-flow/SKILL.md.claude/skills/argent-tv-interact/SKILL.md.codex/config.toml.mcp.json.vscode/mcp.jsonskills-lock.json
…明記・ドキュメント内矛盾を修正 Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Member
Author
CodeRabbit 指摘対応まとめ未解決の 14 件（+ diff 範囲外 1 件）を検証し、13 件を修正、1 件を方針変更のうえ修正、1 件を非該当としてスキップしました。修正はコミット d42c800（ 修正内容:
提案と異なる対応をした 1 件:
スキップした 1 件:
検証: |
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
1 participant
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.
概要
Argent MCP ツールキット（iOS シミュレータ / Android エミュレータ / Chromium アプリ操作用の MCP サーバー）のスキル・エージェント・設定ファイル一式を追加する。アプリ本体のコードには変更なし。
変更の種類
変更内容
.claude/skills/と.agents/skills/の両方に配置）: デバイス操作、iOS/Android セットアップ、Metro デバッガ、React Native プロファイラ・最適化、ネイティブプロファイラ、スクリーンショット diff、UI フローテスト、フロー記録、TV 操作、Argent Lensargent-environment-inspectorサブエージェント定義と品質管理チェックリストを追加（.claude/agents/）.claude/rules/argent.md）.mcp.json、.vscode/mcp.json、.codex/config.toml、.claude/settings.json）skills-lock.json）テスト
npm run lintが通ることnpm testが通ることnpm run typecheckが通ること省略: アプリ本体（
src/など）のコード変更はないため実行対象外。関連Issue
スクリーンショット（任意）
Summary by CodeRabbit
新機能
ドキュメント