zhangjies-skills-and-mcp

# Zhangjie's Skills & MCP for Claude Code One-click replication of Zhangjie's Claude Code full configuration: 22 custom skills, 20 Slash Commands, MCP Server, automatic trigger rules, writing style learning system, and more. ## One-Click Installation ```bash git clone https://github.com/zjie64821-ui/zhangjies-skills-and-mcp.git cd zhangjies-skills-and-mcp chmod +x install.sh ./install.sh ``` The installation script will automatically copy all files to `~/.claude/` and place the configuration files during the first installation (without overwriting existing configurations). ## Post-Installation Requirements ### 1. Replace API Key Edit `~/.claude/settings.json`: ```json { "env": { "ANTHROPIC_AUTH_TOKEN": "Your API Key", "ANTHROPIC_BASE_URL": "Your API proxy address (or delete this line to use the official API)" } } ``` Edit `~/.claude/mcp.json`: ```json { "GLM at https://open.bigmodel.cn/" } ``` ### 2. Install Other MCP Servers (Highly Recommended) These MCP servers provide core functionality and need to be installed separately: | MCP Server | Installation Command | Purpose | |---|---|---| | **Chrome DevTools** | `npx @anthropic-ai/chrome-devtools-mcp@latest` | Browser control, external brain search | | **Zotero** | `npx zotero-mcp@latest` | Literature management | | **ArXiv** | `pip install arxiv-mcp-server` | Paper search | | **Tavily** | `npx tavily-mcp@latest` | Deep web search | | **Memory** | `npx @anthropic-ai/memory-mcp@latest` | Knowledge graph memory | | **Sequential Thinking** | `npx @anthropic-ai/sequential-thinking-mcp@latest` | Structured reasoning | | **OpenAlex** | `pip install openalex-mcp` | Academic search | | **Synapse (Obsidian)** | `npx synapse-mcp@latest` | Obsidian knowledge base | Run `/mcp` in Claude Code to add servers after installation. ### 3. Restart Claude Code ## Included Content ### Slash Commands (20) | Command | Function | |---|---| | `/xlsx` | Excel spreadsheet processing | | `/research-master` | PDF literature deep analysis (V4.5) | | `/content-writer` | Content research and writing assistant | | `/changelog` | Git commit to Changelog | | `/pptx` | PowerPoint presentation processing | | `/style-learn` | Writing style learning engine | | `/obsidian` | Obsidian note library control | | `/mcp-build` | MCP server build guide | | `/research-agent` | Academic research automatic production line (V4.0) | | `/zotero` | Zotero literature management | | `/docx` | Word document processing (including format cloning) | | `/plan` | File-based planning | | `/file-organize` | File organization assistant | | `/pdf` | PDF processing (merging, splitting, extracting, OCR) | | `/cnki-search` | CNKI search | | `/youtube-dl` | YouTube video download | | `/style-write` | Style-driven writing engine | | `/github-sync` | GitHub synchronization (LFS support) | | `/gs-search` | Google Scholar search | | `/webapp-test` | Web application testing (Playwright) | ### Skills (22) In addition to the above 20 commands: - **Data Pipeline**: Python data processing pipeline engine - **Stata**: Econometrics analysis engine - **Policy Eval**: LLM policy text evaluation engine - **Lit Download**: Academic literature batch download and parsing engine ### MCP Server - **GLM-4V Image Analysis**: Custom MCP Server, calling GLM-4V-Flash model to analyze local images ### Core Rules (CLAUDE.md) - **Automatic Image Analysis**: Detect images and analyze immediately without user request - **Writing Style Learning**: Automatically check/learn style before writing tasks, AI flavor removal rules - **MCP Automatic Triggering**: ArXiv, Memory, Tavily, etc. automatically activated by keywords - **Skill Automatic Triggering**: Stata, DataPipeline, PolicyEval triggered by semantics - **External Brain Search**: Specify Grok/Gemini/Xiaohongshu/Bilibili to open automatically - **Chrome DevTools Exclusive Browser**: HBW Launcher management, Google Scholar mirroring ### Other Components - **HBW Browser Launcher** (`~/hbw_launcher.sh`): Chrome DevTools exclusive browser launcher - **SessionStart Hook**: Automatically start browser - **Permission Whitelist**: Pre-configured common tool permissions, reducing permission pop-ups ## Directory Structure ``` zhangjies-skills-and-mcp/ ├── commands/ # 20 Slash Commands ├── skills/ # 22 Skills (including Python scripts) ├── glm-mcp-server/ # GLM-4V Image Analysis MCP Server ├── scripts/ # HBW Launcher and other auxiliary scripts ├── configs/ # Configuration file templates (need to replace API Key) │ ├── settings.json │ ├── settings.local.json │ ├── mcp.json │ └── CLAUDE.md ├── install.sh # One-click installation script └── README.md ``` ## Custom Model Configuration The default configuration uses Volces/Ark API proxy, mapping as follows: - Opus → kimi-k2.6 - Sonnet → minimax-m2.7 - Haiku → glm-5.1 To use Anthropic official API, delete `ANTHROPIC_BASE_URL` and `ANTHROPIC_AUTH_TOKEN`. ## License