KnowledgeSmith MCP
Enables efficient editing of RBT documents with structured operations that read and modify specific sections or blocks. Reduces LLM token consumption by 80-95% compared to full file operations through smart caching and partial document access.
README
RBT MCP Server
MCP Server for editing RBT documents with partial operations. Reduces token consumption by 80-95% compared to full file read/write operations.
🎯 Key Features
- Token Optimization: 80-95% reduction in LLM token usage
- Structured Operations: Edit specific sections/blocks without loading entire documents
- Smart Caching: LRU + TTL cache for frequently accessed documents
- TASK Fuzzy Search: Find TASK files by index (e.g., "001" matches "TASK-001-PathResolver.md")
- Template-based Creation: Auto-fill placeholders (project-id, feature-id, date)
- 13 MCP Tools: Complete CRUD operations for RBT documents
📦 Installation
Option 1: Install from source (uv)
# Clone repository
git clone https://github.com/yourusername/KnowledgeSmith.git
cd KnowledgeSmith
# Install with uv
uv pip install -e .
Option 2: Direct installation
uv pip install rbt-mcp-server
🚀 Quick Start
1. Configure Claude Desktop
Add to ~/Library/Application Support/Claude/claude_desktop_config.json:
{
"mcpServers": {
"rbt-document-editor": {
"type": "stdio",
"command": "rbt-mcp-server",
"env": {
"RBT_ROOT_DIR": "/path/to/your/documents"
}
}
}
}
Or use full uv command:
{
"mcpServers": {
"rbt-document-editor": {
"type": "stdio",
"command": "uv",
"args": ["run", "rbt-mcp-server"],
"env": {
"RBT_ROOT_DIR": "/path/to/your/documents"
}
}
}
}
2. Set Environment Variable
export RBT_ROOT_DIR=/path/to/your/documents
3. Test the Server
rbt-mcp-server
📚 Available MCP Tools
- get_outline - Get document structure (saves 80% tokens)
- read_content - Read specific section/block (saves 90% tokens)
- update_info - Update status/dependencies
- update_section_summary - Update section summary
- create_section - Create new sub-section
- create_block - Create paragraph/code/list/table
- update_block - Update block content
- delete_block - Delete block
- append_list_item - Add item to list
- update_table_row - Update table row
- append_table_row - Add table row
- create_document - Create from template
- clear_cache - Clear document cache
See rbt_mcp_server/README.md for detailed usage.
📊 Token Savings
| Operation | Traditional | MCP | Savings |
|---|---|---|---|
| Read structure | 4,000 | 800 | 80% |
| Update status | 8,000 | 300 | 96% |
| Add list item | 8,000 | 1,000 | 88% |
| Complete TASK | 44,000 | 3,000 | 93% |
🧪 Development
Install development dependencies:
uv sync --dev
Run tests:
RBT_ROOT_DIR=/test/root uv run pytest -v
Test coverage:
RBT_ROOT_DIR=/test/root uv run pytest --cov=rbt_mcp_server --cov-report=html
📝 License
MIT License
🤝 Contributing
Contributions welcome! Please open an issue or submit a pull request.
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。