Back to Details

claude-config-optimizer

Compare original and translation side by side

🇺🇸

Original

English
🇨🇳

Translation

Chinese

Claude Config Optimizer Skill

Claude配置优化器技能

Rules and tracking for Claude Code configuration optimization.
用于Claude Code配置优化的规则与跟踪机制。

1. Config File Location

1. 配置文件位置

User's Claude Code config is stored at: @/ghq/github.com/i9wa4/dotfiles/config/agents/ @/ghq/github.com/i9wa4/dotfiles/config/claude/
Key files:
  • settings.json
    - Main settings
  • CLAUDE.md
    - Persona and core rules
  • rules/
    - Detailed rules (auto-loaded at startup)
  • skills/
    - Specialized knowledge (loaded on demand)
  • agents/
    - Custom agents
  • commands/
    - Slash commands
用户的Claude Code配置存储在以下路径： @/ghq/github.com/i9wa4/dotfiles/config/agents/ @/ghq/github.com/i9wa4/dotfiles/config/claude/
关键文件：
  • settings.json
    - 主配置文件
  • CLAUDE.md
    - 角色设定与核心规则
  • rules/
    - 详细规则（启动时自动加载）
  • skills/
    - 专业知识（按需加载）
  • agents/
    - 自定义Agent
  • commands/
    - 斜杠命令

2. Fetch CHANGELOG

2. 获取CHANGELOG

Use 
gh
command to fetch the latest CHANGELOG:
sh
gh api repos/anthropics/claude-code/contents/CHANGELOG.md \
  --jq '.content' | base64 -d > .i9wa4/tmp/claude-code-changelog.md
Then read the file to analyze.
使用
gh
命令获取最新的CHANGELOG：
sh
gh api repos/anthropics/claude-code/contents/CHANGELOG.md \
  --jq '.content' | base64 -d > .i9wa4/tmp/claude-code-changelog.md
随后读取该文件进行分析。

3. CHANGELOG Operations

3. CHANGELOG操作

3.1. Latest Release Summary

3.1. 最新版本发布摘要

  1. Fetch CHANGELOG using the command above
  2. Extract the first 
    ## x.x.x
    section
  3. Categorize changes into:
    • New features (Added)
    • Bug fixes (Fixed)
    • Improvements (Improved/Changed)
    • Deprecations (Deprecated)
  4. Present in Japanese with brief explanations
  1. 使用上述命令获取CHANGELOG
  2. 提取第一个
    ## x.x.x
    章节
  3. 将变更分类为：
    • 新功能（Added）
    • Bug修复（Fixed）
    • 改进（Improved/Changed）
    • 弃用（Deprecated）
  4. 以日语呈现并附带简要说明

3.2. Version Diff

3.2. 版本差异对比

  1. Ask user for start and end versions
  2. Extract all sections between those versions
  3. Summarize cumulative changes
  4. Highlight breaking changes and deprecations
  1. 询问用户起始版本和结束版本
  2. 提取这两个版本之间的所有章节
  3. 汇总累计变更
  4. 突出显示破坏性变更和弃用内容

3.3. Breaking Changes Detection

3.3. 破坏性变更检测

  1. Search for keywords: 
    Deprecated
    , 
    Removed
    , 
    Breaking
    , 
    Changed
  2. List affected settings and migration paths
  3. Check user's config for affected settings
  1. 搜索关键词：
    Deprecated
    、
    Removed
    、
    Breaking
    、
    Changed
  2. 列出受影响的配置项和迁移路径
  3. 检查用户配置中是否存在受影响的设置

4. Specification Reference

4. 规范参考

For detailed questions about Claude Code specifications, features, and usage:
  • YOU MUST: Use the 
    claude-code-guide
    subagent via Task tool
  • Example queries: hooks, MCP servers, settings, IDE integrations
text
Task tool with subagent_type: claude-code-guide
对于Claude Code规范、功能和使用的详细问题：
  • 必须：通过Task工具调用
    claude-code-guide
    子Agent
  • 示例查询：hooks、MCP服务器、设置、IDE集成
text
Task tool with subagent_type: claude-code-guide

5. Settings Categories

5. 配置分类

CategoryExamples
Display
showTurnDuration
, 
language
Behavior
respectGitignore
, 
autoUpdate
Tools
disallowedTools
, 
allowedTools
MCPMCP server configurations
HooksPreToolUse, PostToolUse, Stop hooks
Plans
plansDirectory
分类示例
显示
showTurnDuration
、
language
行为
respectGitignore
、
autoUpdate
工具
disallowedTools
、
allowedTools
MCPMCP服务器配置
HooksPreToolUse、PostToolUse、Stop hooks
计划
plansDirectory

6. CLAUDE.md Design Guidelines

6. CLAUDE.md设计指南

  • YOU MUST: Focus only on persona and core guidelines
  • YOU MUST: Split detailed rules into 
    rules/
  • NEVER: Include unnecessary information at startup (reference links, usage details)
  • 必须：仅聚焦于角色设定和核心准则
  • 必须：将详细规则拆分至
    rules/
    目录
  • 禁止：在启动时包含不必要的信息（参考链接、使用细节）

7. Configuration Usage

7. 配置使用方式

TypeLoad TimingPurpose
CLAUDE.md / rules/Full load at startupGlobal rules always applied
commands/Explicit user invocationPredefined prompts, workflows
skills/Auto-triggered by conversationSpecialized knowledge
agents/Delegated via Task toolIndependent context
类型加载时机用途
CLAUDE.md / rules/启动时完整加载始终应用的全局规则
commands/用户显式调用预定义提示词、工作流
skills/对话时自动触发专业知识
agents/通过Task工具委托独立上下文

8. Optimization Checklist

8. 优化检查清单

8.1. CLAUDE.md Review

8.1. CLAUDE.md审核

Check the following when editing CLAUDE.md:
  • Is the persona definition concise?
  • Are basic rules truly needed at all times?
  • Can detailed explanations be moved to rules/ or skills/?
  • Have reference links been moved to skills?
编辑CLAUDE.md时需检查以下内容：
  • 角色定义是否简洁？
  • 基础规则是否确实需要始终生效？
  • 详细说明是否可以移至rules/或skills/？
  • 参考链接是否已移至skills？

8.2. Permission System Review

8.2. 权限系统审核

Check settings.json permissions block:
  • Are deny rules using modern syntax 
    Bash(cmd *)
    not deprecated 
    :*
    ?
  • Are critical commands blocked (git push, git rebase, git reset, rm, sudo)?
  • Is 
    git -C *
    blocked to prevent cross-repo operations?
  • Is defaultMode appropriate ("plan" or "dontAsk")?
  • Are sensitive paths blocked (secrets, .env, .ssh, keys, tokens)?
  • Are allow rules necessary or can defaultMode handle it?
检查settings.json中的权限块：
  • 拒绝规则是否使用现代语法
    Bash(cmd *)
    而非已弃用的
    :*
    ？
  • 关键命令是否被阻止（git push、git rebase、git reset、rm、sudo）？
  • git -C *
    是否被阻止以防止跨仓库操作？
  • defaultMode是否合适（"plan"或"dontAsk"）？
  • 敏感路径是否被阻止（secrets、.env、.ssh、keys、tokens）？
  • 允许规则是否必要，或者defaultMode是否可以处理？

9. File Structure Maintenance

9. 文件结构维护

When adding/removing files in rules/, skills/, agents/, or commands/:
  • YOU MUST: Update corresponding table in CLAUDE.md section 4
  • YOU MUST: Keep tables alphabetically sorted or logically grouped
  • IMPORTANT: Verify actual files match documentation after changes
在rules/、skills/、agents/或commands/中添加/删除文件时：
  • 必须：更新CLAUDE.md第4节中的对应表格
  • 必须：保持表格按字母顺序排序或逻辑分组
  • 重要：变更后验证实际文件是否与文档一致

10. Optimization Tracking

10. 优化跟踪

Last reviewed Claude Code version: v2.1.37 (2026-02-09)
最近审核的Claude Code版本：v2.1.37（2026-02-09）

10.1. Applied Optimizations

10.1. 已应用的优化

  • Persona definition minimized
  • Rules split into rules/ directory
  • Skills split into skills/ directory
  • Agents split into agents/ directory
  • Commands split into commands/ directory
  • Reference links moved to skills
  • language
    setting - set to "follow CLAUDE.md's instructions"
  • mcpToolSearch
    setting - set to "auto:1"
  • ENABLE_TOOL_SEARCH
    env - set to "true" (force enable)
  • plansDirectory
    setting - set to ".i9wa4/plans"
  • Permission system reviewed - sandbox bypass fix confirmed (v2.1.34)
  • Permission deny rules migrated - deprecated 
    :*
    to modern 
    *
    syntax
  • Fast mode - available for Opus 4.6 (v2.1.36)
  • Agent teams - enabled via 
    CLAUDE_CODE_EXPERIMENTAL_AGENT_TEAMS
  • Automatic memory - enabled by default (v2.1.32)
  • 角色定义已最小化
  • 规则拆分至rules/目录
  • 技能拆分至skills/目录
  • Agent拆分至agents/目录
  • 命令拆分至commands/目录
  • 参考链接已移至skills
  • language
    设置 - 设为"follow CLAUDE.md's instructions"
  • mcpToolSearch
    设置 - 设为"auto:1"
  • ENABLE_TOOL_SEARCH
    环境变量 - 设为"true"（强制启用）
  • plansDirectory
    设置 - 设为".i9wa4/plans"
  • 权限系统已审核 - 沙箱绕过修复已确认（v2.1.34）
  • 权限拒绝规则已迁移 - 从已弃用的
    :*
    语法改为现代
    *
    语法
  • 快速模式 - 对Opus 4.6可用（v2.1.36）
  • Agent团队 - 通过
    CLAUDE_CODE_EXPERIMENTAL_AGENT_TEAMS
    启用
  • 自动记忆 - 默认启用（v2.1.32）

10.2. Pending Considerations

10.2. 待考虑事项

  • SQL schema validation - moved to databricks skill (Section 8)
  • TeammateIdle/TaskCompleted hooks - for future agent workflow automation
  • Agent memory frontmatter - 
    memory: user|project|local
    for stateful agents
  • Task(agent_type) restrictions - enhanced security for subagent spawning
  • SQL模式验证 - 移至databricks技能（第8节）
  • TeammateIdle/TaskCompleted hooks - 用于未来的Agent工作流自动化
  • Agent记忆前置元数据 - 
    memory: user|project|local
    用于有状态Agent
  • Task(agent_type)限制 - 增强子Agent生成的安全性

10.3. Not Adopting

10.3. 不采用的配置

  • showTurnDuration
    - keep default (show duration)
  • reducedMotionMode
    - keep default (animations enabled)
  • Keybindings customization - use defaults
  • Completion status reporting - no consumer for this output
  • showTurnDuration
    - 保留默认值（显示时长）
  • reducedMotionMode
    - 保留默认值（启用动画）
  • 快捷键自定义 - 使用默认设置
  • 完成状态报告 - 无此输出的消费者

10.4. Version Notes

10.4. 版本说明

  • v2.1.37: /fast immediately available after /extra-usage
  • v2.1.36: Fast mode for Opus 4.6
  • v2.1.34: Sandbox bypass vulnerability fixed (security), agent teams crash fix
  • v2.1.33: TeammateIdle/TaskCompleted hooks, agent memory frontmatter, Task(agent_type) restrictions, tmux agent sessions fix
  • v2.1.32: Claude Opus 4.6 available, automatic memory, agent teams research preview, skills auto-load from --add-dir, Bash template literal fix
  • v2.1.31: PDF lock fix, sandbox error fix, system prompt improvement for dedicated tools
  • v2.1.30: Read tool 
    pages
    param for PDFs, 
    /debug
    command, MCP OAuth pre-configured credentials, reduced motion mode
  • v2.1.29: SessionStart hook saved_hook_context performance fix
  • v2.1.27: --from-pr flag, auto PR-session linking, debug log improvements
  • v2.1.23: spinnerVerbs setting, terminal rendering perf, mTLS/proxy fix
  • v2.1.21: Zenkaku number input (JP IME), auto-compact timing fix, task ID reuse fix
  • v2.1.20: PR review status indicator, --add-dir CLAUDE.md loading, TaskUpdate delete, config backup rotation, background agent permission prompt
  • v2.1.19: CLAUDE_CODE_ENABLE_TASKS env, $ARGUMENTS[0] syntax, skill approval change
  • v2.1.18: Customizable keybindings (/keybindings command)
  • v2.1.16: New task management system with dependency tracking
  • v2.1.15: npm deprecation notice, React Compiler UI perf, MCP stdio timeout fix
  • v2.1.14: Bash history autocomplete, plugin pinning to git SHA, memory fixes
  • v2.1.10: Setup hook, OAuth URL copy shortcut, heredoc template literal fix
  • v2.1.9: plansDirectory, auto:N for MCP, PreToolUse additionalContext
  • v2.1.7: keybindings.json, showTurnDuration, MCP tool search auto mode
  • v2.1.3: Merged slash commands and skills
  • v2.1.0: language setting, skill hot-reload, context: fork
  • v2.1.37：/fast命令可在/extra-usage后立即使用
  • v2.1.36：Opus 4.6支持快速模式
  • v2.1.34：修复沙箱绕过漏洞（安全）、修复Agent团队崩溃问题
  • v2.1.33：新增TeammateIdle/TaskCompleted hooks、Agent记忆前置元数据、Task(agent_type)限制、修复tmux Agent会话问题
  • v2.1.32：Claude Opus 4.6可用、自动记忆、Agent团队研究预览、从--add-dir自动加载技能、修复Bash模板字面量问题
  • v2.1.31：修复PDF锁定问题、修复沙箱错误、改进专用工具的系统提示
  • v2.1.30：Read工具支持PDF的
    pages
    参数、新增/debug命令、MCP OAuth预配置凭据、减少动画模式
  • v2.1.29：修复SessionStart hook的saved_hook_context性能问题
  • v2.1.27：新增--from-pr标志、自动PR会话关联、改进调试日志
  • v2.1.23：新增spinnerVerbs设置、优化终端渲染性能、修复mTLS/代理问题
  • v2.1.21：支持全角数字输入（日语输入法）、修复自动压缩计时问题、修复任务ID重用问题
  • v2.1.20：PR审核状态指示器、--add-dir加载CLAUDE.md、删除TaskUpdate、配置备份轮换、后台Agent权限提示
  • v2.1.19：新增CLAUDE_CODE_ENABLE_TASKS环境变量、$ARGUMENTS[0]语法、技能审批变更
  • v2.1.18：可自定义快捷键（/keybindings命令）
  • v2.1.16：新增带依赖跟踪的任务管理系统
  • v2.1.15：npm弃用通知、React Compiler UI性能优化、修复MCP标准输入输出超时问题
  • v2.1.14：Bash历史自动补全、插件固定到git SHA、修复记忆问题
  • v2.1.10：新增Setup hook、OAuth URL复制快捷方式、修复heredoc模板字面量问题
  • v2.1.9：新增plansDirectory、MCP的auto:N模式、PreToolUse的additionalContext
  • v2.1.7：新增keybindings.json、showTurnDuration、MCP工具搜索自动模式
  • v2.1.3：合并斜杠命令与技能
  • v2.1.0：新增language设置、技能热重载、context: fork

11. Response Format (CHANGELOG)

11. CHANGELOG响应格式

text
undefined
text
undefined

Claude Code vX.X.X

Claude Code vX.X.X

New Features

New Features

Bug Fixes

Bug Fixes

  • [Fix description]
  • [Fix description]

Improvements

Improvements

  • [Improvement description]
  • [Improvement description]

Deprecated/Removed

Deprecated/Removed

  • [Affected settings and migration path]
Source: https://github.com/anthropics/claude-code
undefined
  • [Affected settings and migration path]
Source: https://github.com/anthropics/claude-code
undefined

12. site2skill Usage

12. site2skill使用方法

Convert documentation websites into Claude Agent Skills.
Requirements: Python 3.10+, wget (
brew install wget
)
sh
uvx --from git+https://github.com/laiso/site2skill site2skill <URL> <SKILL_NAME>
将文档网站转换为Claude Agent技能。
要求：Python 3.10+、wget（
brew install wget
）
sh
uvx --from git+https://github.com/laiso/site2skill site2skill <URL> <SKILL_NAME>

Example

示例

uvx --from git+https://github.com/laiso/site2skill site2skill https://docs.pay.jp/v1/ payjp

Options:

- `--output, -o` - Output directory (default: `.claude/skills`)
- `--skill-output` - Where to save .skill file (default: current directory)
- `--skip-fetch` - Skip download (reuse existing files)
- `--clean` - Remove temporary files after completion

To update existing skill docs, re-run without `--skip-fetch`.
uvx --from git+https://github.com/laiso/site2skill site2skill https://docs.pay.jp/v1/ payjp

选项：

- `--output, -o` - 输出目录（默认：`.claude/skills`）
- `--skill-output` - .skill文件保存路径（默认：当前目录）
- `--skip-fetch` - 跳过下载（重用现有文件）
- `--clean` - 完成后删除临时文件

要更新现有技能文档，重新运行时不添加`--skip-fetch`参数即可。

13. Reference Links

13. 参考链接

Official Documentation:
Community Resources:
官方文档：
社区资源：

14. Permission System Reference

14. 权限系统参考

14.1. Permission Modes

14.1. 权限模式

ModeDescription
default
Prompts for permission on first use of each tool
acceptEdits
Auto-accepts file edit permissions for the session
plan
Plan Mode: analyze only, no modifications
dontAsk
Auto-denies unless pre-approved via allow rules
bypassPermissions
Skips all prompts (use only in isolated environments)
模式描述
default
首次使用每个工具时提示获取权限
acceptEdits
会话期间自动接受文件编辑权限
plan
仅分析，不进行修改
dontAsk
自动拒绝，除非通过允许规则预先批准
bypassPermissions
跳过所有提示（仅在隔离环境中使用）

14.2. Rule Evaluation Order

14.2. 规则评估顺序

Rules are evaluated: deny -> ask -> allow. First matching rule wins.
规则评估顺序：拒绝 -> 询问 -> 允许。第一个匹配的规则生效。

14.3. Bash Wildcard Patterns

14.3. Bash通配符模式

json
{
  "permissions": {
    "allow": [
      "Bash(npm run *)",
      "Bash(git commit *)",
      "Bash(* --version)",
      "Bash(* --help *)"
    ],
    "deny": ["Bash(git push *)"]
  }
}
NOTE: Space before 
*
matters: 
Bash(ls *)
matches 
ls -la
but not 
lsof
.
json
{
  "permissions": {
    "allow": [
      "Bash(npm run *)",
      "Bash(git commit *)",
      "Bash(* --version)",
      "Bash(* --help *)"
    ],
    "deny": ["Bash(git push *)"]
  }
}
注意：
*
前的空格很重要：
Bash(ls *)
匹配
ls -la
但不匹配
lsof
。

14.4. Read/Edit Path Patterns

14.4. 读取/编辑路径模式

PatternMeaningExample
//path
Absolute path from root
Read(//Users/alice/**)
~/path
Path from home directory
Read(~/.zshrc)
/path
Relative to settings file
Edit(/src/**/*.ts)
path
Relative to current directory
Read(*.env)
NOTE: 
*
matches single directory, 
**
matches recursively.
模式含义示例
//path
从根目录开始的绝对路径
Read(//Users/alice/**)
~/path
从用户主目录开始的路径
Read(~/.zshrc)
/path
相对于配置文件的路径
Edit(/src/**/*.ts)
path
相对于当前目录的路径
Read(*.env)
注意：
*
匹配单个目录，
**
递归匹配所有目录。

14.5. MCP and Task Permissions

14.5. MCP与Task权限

json
{
  "permissions": {
    "allow": ["mcp__puppeteer__*"],
    "deny": ["Task(Explore)"]
  }
}
json
{
  "permissions": {
    "allow": ["mcp__puppeteer__*"],
    "deny": ["Task(Explore)"]
  }
}

14.6. Managed Settings Locations

14.6. 托管配置位置

PlatformPath
macOS
/Library/Application Support/ClaudeCode/managed-settings.json
Linux/WSL
/etc/claude-code/managed-settings.json
Windows
C:\Program Files\ClaudeCode\managed-settings.json
平台路径
macOS
/Library/Application Support/ClaudeCode/managed-settings.json
Linux/WSL
/etc/claude-code/managed-settings.json
Windows
C:\Program Files\ClaudeCode\managed-settings.json

14.7. Managed-Only Settings

14.7. 仅托管配置项

SettingDescription
disableBypassPermissionsMode
Set to "disable" to prevent bypass mode
allowManagedPermissionRulesOnly
Only managed rules apply
allowManagedHooksOnly
Only managed/SDK hooks allowed
配置项描述
disableBypassPermissionsMode
设置为"disable"以防止绕过模式
allowManagedPermissionRulesOnly
仅应用托管规则
allowManagedHooksOnly
仅允许托管/SDK hooks

15. Insights-Based Recommendations

15. 基于使用分析的建议

Based on usage analysis (55K messages, 4.7K sessions):
基于使用分析（55K条消息，4.7K次会话）：

15.1. Applied

15.1. 已应用

  • Schema validation for DB operations: See databricks skill Section 8
  • 数据库操作的模式验证：参见databricks技能第8节

15.2. Not Applied

15.2. 未应用

  • Completion status reporting: No consumer for this output
  • PreToolUse hook for SQL: Handled by skill guidance instead
  • 完成状态报告：无此输出的消费者
  • SQL的PreToolUse hook：由技能指导处理

16. References

16. 参考资料

  • Hooks Reference: 
    https://code.claude.com/docs/en/hooks
  • Permissions Reference: 
    https://code.claude.com/docs/en/permissions
  • Skills Dynamic Context: 
    https://code.claude.com/docs/en/skills#inject-dynamic-context
  • Vercel AGENTS.md Guide: 
    https://vercel.com/blog/agents-md
  • Hooks参考：
    https://code.claude.com/docs/en/hooks
  • 权限参考：
    https://code.claude.com/docs/en/permissions
  • 技能动态上下文：
    https://code.claude.com/docs/en/skills#inject-dynamic-context
  • Vercel AGENTS.md指南：
    https://vercel.com/blog/agents-md