feat: add Claude Code integration and unified agent settings - #37
Merged
Conversation
📝 WalkthroughWalkthrough项目新增 Claude Code 集成。Codex 与 Claude Code 现在共享智能体适配器、Hook 服务和设置界面，同时保留独立配置、事件、验证状态和操作流程。新增失败事件处理、回合生命周期处理、测试、文档和图标说明。 ChangesClaude Code 集成
Estimated code review effort: 4 (Complex) | ~60 minutes Sequence Diagram(s)sequenceDiagram
participant ClaudeCode
participant hook_server
participant live_status
participant settings
ClaudeCode->>hook_server: 发送智能体事件
hook_server->>live_status: 转发规范化事件
live_status->>settings: 更新运行状态
settings-->>ClaudeCode: 显示连接与事件状态
Possibly related PRs
🚥 Pre-merge checks | ✅ 4 | ❌ 1❌ Failed checks (1 warning)
✅ Passed checks (4 passed)
✨ Finishing Touches 💡 1📝 Generate docstrings 💡
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: 1
🧹 Nitpick comments (7)
src-tauri/src/hook_installer/claude_code.rs (1)
16-22: 📐 Maintainability & Code Quality | 🔵 Trivial | 💤 Low value建议去掉
expect，用显式错误代替。当前分支在
settings_path返回None时调用expect。虽然此处home一定是Some，但该不变量依赖settings_path的内部实现。若后续修改settings_path，此处会 panic。可用ok_or_else返回错误，保持函数的Result契约。♻️ 建议的重构
pub(super) fn spec() -> Result<HookSpec, String> { let config_dir = std::env::var_os("CLAUDE_CONFIG_DIR"); - let path = if let Some(path) = settings_path(config_dir.as_deref(), None) { - path - } else { - settings_path(None, Some(&config::home_dir()?)).expect("home path is present") - }; + let path = match settings_path(config_dir.as_deref(), None) { + Some(path) => path, + None => settings_path(None, Some(&config::home_dir()?)) + .ok_or_else(|| "无法确定 Claude Code 配置路径".to_string())?, + };🤖 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-tauri/src/hook_installer/claude_code.rs` around lines 16 - 22, Update spec’s fallback settings_path handling to remove expect("home path is present") and convert a None result into an explicit Err via ok_or_else, preserving the function’s Result<HookSpec, String> contract and returning a descriptive error instead of panicking.src/settings.ts (3)
391-398: 📐 Maintainability & Code Quality | 🔵 Trivial | ⚡ Quick win建议复用
integrationIds常量。这两处重复写出
["codex", "claude-code"] as const，而 Line 24 已定义integrationIds。若以后新增智能体，需要修改三处。请改为遍历integrationIds。♻️ 建议的改动
- for (const agent of ["codex", "claude-code"] as const) { + for (const agent of integrationIds) { const definition = integrationDefinitions[agent];-for (const agent of ["codex", "claude-code"] as const) { +for (const agent of integrationIds) { const definition = integrationDefinitions[agent];Also applies to: 683-700
🤖 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/settings.ts` around lines 391 - 398, Reuse the existing integrationIds constant in both affected loops instead of repeating the literal ["codex", "claude-code"] as const. Update the loops around integrationConfig(agent) and the corresponding block near the alternate occurrence so adding an agent only requires changing integrationIds.
825-837: 🚀 Performance & Scalability | 🔵 Trivial | ⚡ Quick win建议对
refreshHookStatus增加去抖。事件监听器对每个到达的智能体事件调用
refreshHookStatus。每次调用发起hook_status与hook_runtime_status两个 IPC 请求。后端hook_installer::status会读取并解析该智能体的 Hook 配置文件。在活跃任务期间，
PreToolUse与PostToolUse事件频繁到达。加上 Line 834 的 5 秒定时刷新，磁盘读取会累积。建议对每个智能体加入短去抖窗口（例如 300ms），合并连续事件。🤖 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/settings.ts` around lines 825 - 837, 为 refreshHookStatus 增加按智能体独立的短去抖机制（约 300ms），使 AGENT_EVENT_CHANNEL 监听器和 5 秒 healthTimer 在窗口内的连续刷新请求合并执行；保留事件对 codex 与 claude-code 的现有筛选和最终状态刷新行为，并在 refreshHookStatus 附近复用统一的去抖调度逻辑。
543-550: 📐 Maintainability & Code Quality | 🔵 Trivial | 💤 Low value建议用
displayName替换硬编码的 "Claude Code" 文案。这些分支对任意
agent执行，但文案固定写 "Claude Code"。当前后端只对 Claude Code 返回globallyDisabled: true，所以显示结果正确。该正确性依赖后端实现，属隐式耦合。若以后其他智能体也支持全局禁用，Codex 卡片会显示 Claude Code 文案。建议改用已有的
displayName变量，例如${displayName} 已全局禁用所有 Hooks。同样适用于 Line 747、779、798 的提示文案。Also applies to: 565-569
🤖 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/settings.ts` around lines 543 - 550, Replace the hard-coded “Claude Code” text in the hook status messages assigned to linkControl.title and hookElement.textContent with the existing displayName interpolation, including the corresponding messages at the other referenced status-message branches. Preserve the existing global-disable behavior and wording while making each agent’s UI use its own displayName.e2e/readme-screenshots.screenshot.ts (1)
42-42: 📐 Maintainability & Code Quality | 🔵 Trivial | ⚡ Quick win建议让 mock 按
agent参数返回数据。
hook_statusmock 忽略agent参数，对两个智能体都返回~/.codex/hooks.json与11/11。Claude Code 的真实期望事件数为 13。同时 Line 42 把
claudeCode.hooksEnabled设为false。结合已安装的 mock 状态，refreshHookStatus会让 Claude Code 卡片进入"已暂停"分支（见src/settings.tsLine 570）。请确认 README 截图需要该状态。若需要"已连接"，请把hooksEnabled设为true，并按 agent 返回对应路径与事件数。Also applies to: 80-88
🤖 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 `@e2e/readme-screenshots.screenshot.ts` at line 42, Update the hook_status mock used by the README screenshot to branch on the agent parameter, returning the Codex hook path and 11/11 counts for Codex and the Claude hook path and 13/13 counts for Claude Code. In the claudeCode configuration, set hooksEnabled to true if the screenshot should show the connected state; otherwise confirm and preserve the intended paused state.src-tauri/src/hook_server.rs (1)
374-382: 🚀 Performance & Scalability | 🔵 Trivial | ⚡ Quick win建议缓存验证指纹，减少每事件的磁盘读取。
emit_event对每个真实事件调用hook_verification::record。record会调用verification_fingerprint(agent)，该函数读取并解析该 agent 的 Hook 配置文件，随后record再读取一次验证状态文件。该工作发生在 Hook 接收线程上。本次改动使两个 agent 都进入这条路径，事件量随之上升。
PreToolUse与PostToolUse频率较高，磁盘读取会累积。建议在指纹已验证后跳过重复读取，或按 agent 缓存指纹并按文件 mtime 失效。🤖 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-tauri/src/hook_server.rs` around lines 374 - 382, 优化 emit_event 调用的 hook_verification::record 与 verification_fingerprint 流程：按 agent 缓存已验证的指纹，避免每个真实事件重复读取和解析 Hook 配置及验证状态文件；当配置文件 mtime 变化时使缓存失效并重新验证，保持现有验证结果与错误处理行为。src-tauri/src/hook_verification.rs (1)
41-48: 🔒 Security & Privacy | 🔵 Trivial | ⚡ Quick win建议校验
agent后再拼接文件名。
path把agent直接插入文件名。agent源自 Tauri 命令参数。目前所有调用方（verified_at、record、clear）都先经过hook_installer::verification_fingerprint或hook_installer::install，这两者会对未知 agent 返回错误，因此路径遍历当前不可达。该保护是隐式的。若以后新增直接调用
path的入口，含../或路径分隔符的 agent 会写到配置目录之外。建议在path内做一次显式校验。🛡️ 建议的防御性改动
fn path(agent: &str) -> Result<std::path::PathBuf, String> { + if !matches!(agent, hook_installer::CODEX | hook_installer::CLAUDE_CODE) { + return Err(format!("不支持的 Agent：{agent}")); + } let name = if agent == hook_installer::CODEX {🤖 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-tauri/src/hook_verification.rs` around lines 41 - 48, 在 path 函数内部显式校验 agent，仅允许受支持且不会包含路径遍历片段或路径分隔符的值；对无效 agent 返回错误后再拼接文件名。保留 hook_installer::CODEX 的特殊文件名及其他合法 agent 的现有命名规则。
🤖 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/terminal-event-ledger.ts`:
- Around line 47-50: Update the terminalSession.turnEnded branch in the
event-filtering logic to validate payload.timestamp before allowing
startsSession, UserPromptSubmit, or SessionEnd events without a turnId; reject
stale events while preserving the existing handling for current events. Add
regression coverage in terminal-event-ledger.test.ts for late UserPromptSubmit
and SessionStart events, ensuring they cannot clear the ended session and allow
subsequent stale PostToolUse events.
---
Nitpick comments:
In `@e2e/readme-screenshots.screenshot.ts`:
- Line 42: Update the hook_status mock used by the README screenshot to branch
on the agent parameter, returning the Codex hook path and 11/11 counts for Codex
and the Claude hook path and 13/13 counts for Claude Code. In the claudeCode
configuration, set hooksEnabled to true if the screenshot should show the
connected state; otherwise confirm and preserve the intended paused state.
In `@src-tauri/src/hook_installer/claude_code.rs`:
- Around line 16-22: Update spec’s fallback settings_path handling to remove
expect("home path is present") and convert a None result into an explicit Err
via ok_or_else, preserving the function’s Result<HookSpec, String> contract and
returning a descriptive error instead of panicking.
In `@src-tauri/src/hook_server.rs`:
- Around line 374-382: 优化 emit_event 调用的 hook_verification::record 与
verification_fingerprint 流程：按 agent 缓存已验证的指纹，避免每个真实事件重复读取和解析 Hook
配置及验证状态文件；当配置文件 mtime 变化时使缓存失效并重新验证，保持现有验证结果与错误处理行为。
In `@src-tauri/src/hook_verification.rs`:
- Around line 41-48: 在 path 函数内部显式校验 agent，仅允许受支持且不会包含路径遍历片段或路径分隔符的值；对无效 agent
返回错误后再拼接文件名。保留 hook_installer::CODEX 的特殊文件名及其他合法 agent 的现有命名规则。
In `@src/settings.ts`:
- Around line 391-398: Reuse the existing integrationIds constant in both
affected loops instead of repeating the literal ["codex", "claude-code"] as
const. Update the loops around integrationConfig(agent) and the corresponding
block near the alternate occurrence so adding an agent only requires changing
integrationIds.
- Around line 825-837: 为 refreshHookStatus 增加按智能体独立的短去抖机制（约 300ms），使
AGENT_EVENT_CHANNEL 监听器和 5 秒 healthTimer 在窗口内的连续刷新请求合并执行；保留事件对 codex 与
claude-code 的现有筛选和最终状态刷新行为，并在 refreshHookStatus 附近复用统一的去抖调度逻辑。
- Around line 543-550: Replace the hard-coded “Claude Code” text in the hook
status messages assigned to linkControl.title and hookElement.textContent with
the existing displayName interpolation, including the corresponding messages at
the other referenced status-message branches. Preserve the existing
global-disable behavior and wording while making each agent’s UI use its own
displayName.
🪄 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 Plus
Run ID: 017fd08c-c1f2-4ee8-9618-2e8b6de88db4
⛔ Files ignored due to path filters (2)
assets/agent-icons/claude-code.svgis excluded by!**/*.svgassets/agent-icons/codex.svgis excluded by!**/*.svg
📒 Files selected for processing (29)
README.mdREADME.zh-CN.mdassets/agent-icons/LICENSE.lobe-iconsassets/agent-icons/README.mde2e/readme-screenshots.screenshot.tse2e/settings.e2e.tssrc-tauri/src/config.rssrc-tauri/src/hook_installer.rssrc-tauri/src/hook_installer/claude_code.rssrc-tauri/src/hook_installer/codex.rssrc-tauri/src/hook_installer/unix.rssrc-tauri/src/hook_installer/windows.rssrc-tauri/src/hook_server.rssrc-tauri/src/hook_verification.rssrc-tauri/src/lib.rssrc/agents/claude-code.tssrc/agents/registry.test.tssrc/agents/registry.tssrc/agents/types.tssrc/live-status.test.tssrc/live-status.tssrc/reaction-controller.test.tssrc/reaction-controller.tssrc/settings.htmlsrc/settings.tssrc/styles.csssrc/terminal-event-ledger.test.tssrc/terminal-event-ledger.tssrc/types.ts
Owner
Author
|
@monkeyscan review |
There was a problem hiding this comment.
🧹 Nitpick comments (1)
src/terminal-event-ledger.test.ts (1)
84-87: 🗄️ Data Integrity & Integration | 🔵 Trivial | ⚡ Quick win让测试覆盖
recordActivity的状态转换。这两个用例只调用
shouldIgnore。它们没有模拟事件入口在结果为false时调用recordActivity。因此，测试没有直接验证迟到事件被错误放行后会清除turnEnded，以及后续PostToolUse是否仍被拒绝。请按生产顺序处理事件：调用
shouldIgnore，仅在返回false时调用recordActivity，然后断言后续PostToolUse仍被忽略。建议的最小修改
- expect(ledger.shouldIgnore(claudeEvent("UserPromptSubmit", 1))).toBe(true); + const stalePrompt = claudeEvent("UserPromptSubmit", 1); + const promptIgnored = ledger.shouldIgnore(stalePrompt); + if (!promptIgnored) ledger.recordActivity(stalePrompt); expect(ledger.shouldIgnore(claudeEvent("PostToolUse", 2))).toBe(true); + expect(promptIgnored).toBe(true); - expect(ledger.shouldIgnore(claudeEvent("SessionStart", 1))).toBe(true); + const staleSessionStart = claudeEvent("SessionStart", 1); + const sessionStartIgnored = ledger.shouldIgnore(staleSessionStart); + if (!sessionStartIgnored) ledger.recordActivity(staleSessionStart); expect(ledger.shouldIgnore(claudeEvent("PostToolUse", 2))).toBe(true); + expect(sessionStartIgnored).toBe(true);Also applies to: 99-102
🤖 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 `@src/terminal-event-ledger.test.ts` around lines 84 - 87, Update the test around ledger.shouldIgnore and recordActivity to process events in production order: call shouldIgnore for each event and invoke recordActivity only when it returns false. Assert that the late UserPromptSubmit does not clear the ended-turn state and that the subsequent PostToolUse remains ignored.
🤖 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.
Nitpick comments:
In `@src/terminal-event-ledger.test.ts`:
- Around line 84-87: Update the test around ledger.shouldIgnore and
recordActivity to process events in production order: call shouldIgnore for each
event and invoke recordActivity only when it returns false. Assert that the late
UserPromptSubmit does not clear the ended-turn state and that the subsequent
PostToolUse remains ignored.
ℹ️ Review info
⚙️ Run configuration
Configuration used: Organization UI
Review profile: CHILL
Plan: Pro Plus
Run ID: b3428762-8336-413d-9201-354a2b209ac8
📒 Files selected for processing (2)
src/terminal-event-ledger.test.tssrc/terminal-event-ledger.ts
🚧 Files skipped from review as they are similar to previous changes (1)
- src/terminal-event-ledger.ts
|
@monkeyscan review 下这个 PR |
收到，我正在处理中，完成后会更新结果。 |
|
PR Title: feat: add Claude Code integration and unified agen... Commit: 本次 PR 为 Agent Cat 引入 Claude Code 集成，与现有 Codex 集成并行运行。主要变更包括：
整体评估：架构设计合理，将 Agent 相关逻辑抽象为 |
已触发 PR Review，完成后会在当前 PR 中发布评审结果。 |
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
Claude Code integration
Review fixes
Validation
Summary by CodeRabbit
新功能
文档
测试