claude-codex-settings

# claude-codex-settings My personal Claude [Code](https://github.com/anthropics/claude-code)/[Desktop](https://claude.ai/download) and [OpenAI Codex](https://developers.openai.com/codex) setup with battle-tested commands and MCP servers that I use daily. ## Setup - Install nvm and Node.js (v22+ recommended) ```bash curl -o- https://raw.githubusercontent.com/nvm-sh/nvm/v0.40.3/install.sh | bash \. "$HOME/.nvm/nvm.sh" nvm install 22 node -v # Should print "v22.17.1". nvm current # Should print "v22.17.1". ``` - Install Claude Code and/or OpenAI Codex: ```bash npm install -g @anthropic-ai/claude-code ``` ```bash npm install -g @openai/codex ``` - Optionally install [Claude Code VSCode extension](https://docs.claude.com/en/docs/claude-code/vs-code) and/or [Codex VSCode extension](https://developers.openai.com/codex/ide) for 100% IDE integration. - Install jq (required for hooks): **macOS:** ```bash brew install jq ``` **Ubuntu/Debian:** ```bash sudo apt-get install jq ``` **Other Linux distributions:** ```bash # Check your package manager, e.g.: # sudo yum install jq (RHEL/CentOS) # sudo pacman -S jq (Arch) ``` - Install code quality tools: ```bash # Python formatting (required for Python hook) pip install ruff docformatter # Prettier for JS/TS/CSS/JSON/YAML/HTML/Markdown/Shell formatting (required for prettier hooks) npm install -g prettier@3.6.2 prettier-plugin-sh ``` - Create a shared guidance symlink for other agents ([agents.md](https://agents.md/)): ```bash ln -s CLAUDE.md AGENTS.md ``` This lets tools like Claude Code, Codex, Gemini CLI, Cursor, and Copilot reuse the same instructions. ## MCP Servers The MCP (Model Context Protocol) configuration lives in [`mcp.json`](./mcp.json). These are some solid MCP server repos worth checking out: - [Azure MCP](https://github.com/Azure/azure-mcp) - 40+ Azure tools (100% free) - [Context7](https://github.com/upstash/context7) - Up-to-date documentation context for 20K+ libraries (100% free) - [GitHub MCP Server](https://github.com/github/github-mcp-server) - 50+ GitHub tools (100% free) - [MongoDB MCP](https://github.com/mongodb-js/mongodb-mcp-server) - Tools for interacting with MongoDB (100% free) - [Paper Search MCP](https://github.com/openags/paper-search-mcp) - Search papers across arXiv, PubMed, bioRxiv, Google Scholar, and more (100% free) - [Playwright MCP](https://github.com/microsoft/playwright-mcp) - 30+ browser/web testing tools (100% free) - [Slack MCP Server](https://github.com/ubie-oss/slack-mcp-server) - 10+ Slack tools (100% free) - [Supabase MCP](https://github.com/supabase-community/supabase-mcp) - Database tools for interacting with Supabase (100% free) - [Configuration guide](https://supabase.com/docs/guides/getting-started/mcp#step-2-configure-your-ai-tool) - [Tavily MCP](https://github.com/tavily-ai/tavily-mcp) - 4 tools for web search and scraping. Better than Claude Code's built-in WebFetch tool (free tier: 1000 monthly requests) OpenAI Codex compatible version of MCP server configurations can be found in [`~/.codex/config.toml`](./config.toml). ## Configuration Claude Code configuration is stored in [`.claude/settings.json`](./.claude/settings.json) and includes: - Model selection (currently using OpusPlan with claude-opus-4-1-20250805 and Sonnet with claude-sonnet-4-5-20250929 - see [model configuration docs](https://docs.anthropic.com/en/docs/claude-code/model-config#opusplan-model-setting)) - Environment variables for optimal Claude Code behavior - Settings for disabling telemetry and non-essential features - Custom hooks for enhancing tool functionality OpenAI Codex configuration is stored in [`~/.codex/config.toml`](./config.toml) and includes: - Default `gpt-5-codex` model with `model_reasoning_effort` set to "high" and served through the Azure `responses` API surface - Azure provider metadata (`model_providers.azure`) with the project-specific base URL and `env_key` secret for authentication ## Statusline The setup includes a custom statusline powered by [ccusage](https://ccusage.com/) that displays Claude usage statistics in real-time. The statusline configuration provides: - **Real-time usage tracking**: Monitor token consumption and API costs as you work - **Offline support**: Cached data ensures statusline works without internet connectivity - **Customizable refresh**: Updates every 2 seconds for responsive feedback - **Turkish localization**: Displays costs and dates in Turkish format (configurable) ### Configuration The statusline is configured through two files: - **[`.claude/settings.json`](./.claude/settings.json)**: Contains the statusline command configuration - **[`.claude/ccusage.json`](./.claude/ccusage.json)**: ccusage-specific settings for locale, timezone, and refresh behavior For detailed setup instructions and customization options, see the [ccusage statusline guide](https://ccusage.com/guide/statusline). ## Agents Specialized agents that run automatically to enhance code quality, stored in [`.claude/agents/`](./.claude/agents/): - [`code-simplifier.md`](./.claude/agents/code-simplifier.md) - Contextual pattern analyzer that ensures new code follows existing project conventions (imports, naming, function signatures, class patterns). Auto-triggers after TodoWrite to maintain codebase consistency. - [`commit-orchestrator.md`](./.claude/agents/commit-orchestrator.md) - Git commit expert that analyzes staged changes, creates optimal commit strategies, and executes commits with meaningful messages. Handles documentation updates and multi-commit scenarios. For more details, see the [Claude Code sub-agents documentation](https://docs.anthropic.com/en/docs/claude-code/sub-agents). ## Hooks Custom hooks that enhance tool usage, configured in [`.claude/settings.json`](./.claude/settings.json): ### Setup Make hook scripts executable after cloning: ```bash chmod +x ./.claude/hooks/*.py ``` ### Web Content Enhancement Hooks These hooks redirect native Claude Code web tools to faster and more reliable Tavily alternatives. Native WebSearch/WebFetch tools take 20-30 seconds while Tavily equivalents complete in 1-2 seconds. Additionally, native WebFetch often fails on bot-protected websites while Tavily can bypass these protections. - **[hook_webfetch_to_tavily_extract.py](./.claude/hooks/hook_webfetch_to_tavily_extract.py)**: Blocks WebFetch and suggests using Tavily extract with advanced depth - **[hook_tavily_extract_to_advanced.py](./.claude/hooks/hook_tavily_extract_to_advanced.py)**: Enhances tavily-extract calls with advanced extraction depth for better content parsing - **[hook_websearch_to_tavily_search.py](./.claude/hooks/hook_websearch_to_tavily_search.py)**: Blocks WebSearch and suggests using Tavily search instead ### Code Quality Hooks Comprehensive auto-formatting system that covers all major file types, designed to eliminate formatting inconsistencies and reduce CI formatting noise. - **Whitespace Cleanup** ([settings.json#L64-L74](./.claude/settings.json#L64-L74)): Automatically removes whitespace from empty lines in Python, JavaScript, and TypeScript files (`.py`, `.js`, `.jsx`, `.ts`, `.tsx`) after any Edit, MultiEdit, Write, or Task operation. Works cross-platform (macOS and Linux). - **Python Code Quality** ([hook_python_code_quality.py](./.claude/hooks/hook_python_code_quality.py)): Automatically formats and lints Python files using ruff and docformatter after Edit/Write/MultiEdit operations. Matches [Ultralytics Actions](https://github.com/ultralytics/actions) pipeline exactly: - `ruff format --line-length 120` - `ruff check --fix --unsafe-fixes --extend-select I,D,UP --target-version py38` - `docformatter --wrap-summaries 120 --wrap-descriptions 120` - **Prettier Formatting** ([hook_prettier_formatting.py](./.claude/hooks/hook_prettier_formatting.py)): Auto-formats JavaScript, TypeScript, CSS, JSON, YAML, HTML, Vue, and Svelte files using prettier. Skips lock files and model.json to prevent conflicts. - **Markdown Formatting** ([hook_markdown_formatting.py](./.claude/hooks/hook_markdown_formatting.py)): Formats Markdown files with prettier, applying special tab-width 4 handling for documentation directories (matches [Ultralytics Actions](https://github.com/ultralytics/actions) docs formatting). - **Bash/Shell Formatting** ([hook_bash_formatting.py](./.claude/hooks/hook_bash_formatting.py)): Formats shell scripts (`.sh`, `.bash`) using prettier-plugin-sh for consistent bash scripting style. - **Ripgrep Enforcement** ([hook_enforce_rg_over_grep.py](./.claude/hooks/hook_enforce_rg_over_grep.py)): Blocks grep and find commands in Bash tool calls, suggesting rg (ripgrep) alternatives for better performance and more features. #### Zero CI Formatting This comprehensive formatting setup is designed to achieve **zero auto-formatting** from CI workflows like [Ultralytics Actions](https://github.com/ultralytics/actions). The hooks cover 95% of typical formatting needs: - ✅ Python (ruff + docformatter) - ✅ JavaScript/TypeScript (prettier) - ✅ CSS/SCSS/Less (prettier) - ✅ JSON/YAML (prettier) - ✅ HTML/Vue/Svelte (prettier) - ✅ Markdown (prettier with docs handling) - ✅ Shell scripts (prettier-plugin-sh) - ✅ Whitespace cleanup All hooks gracefully degrade when tools aren't available, never disrupting Claude Code operations. Python formatting configuration inspired by [onuralpszr's setup](https://github.com/onuralpszr/onuralpszr/blob/main/configs/git-hooks/pre-commit-line-120). For more details, see the [Claude Code hooks documentation](https://docs.anthropic.com/en/docs/claude-code/hooks). ## Commands Custom Claude Code commands that make life easier, stored in [`.claude/commands/`](./.claude/commands/): - [`apply-thinking-to.md`](./.claude/commands/apply-thinking-to.md) - Prompt enhancement using advanced thinking patterns (inspired by [centminmod's version](https://github.com/centminmod/my-claude-code-setup/blob/master/.claude/commands/anthropic/apply-thinking-to.md)) - [`commit-staged-changes.md`](./.claude/commands/commit-staged-changes.md) - Automated commit creation with conventional commit messages - [`explain-architecture-pattern.md`](./.claude/commands/explain-architecture-pattern.md) - Identify and explain architectural patterns and design decisions - [`update-pr-summary.md`](./.claude/commands/update-pr-summary.md) - Generate PR summaries with advanced analytical frameworks ## Extra Resources - [Claude MCP Server Setup](https://docs.anthropic.com/en/docs/claude-code/mcp#project-scope) - [Claude Code Commands Setup](https://docs.anthropic.com/en/docs/claude-code/slash-commands#command-types) ## Thank you for the support! [](https://www.star-history.com/#fcakyon/claude-settings&Date)