Daruka MCP Server
Connects Claude to Daruka project memory, enabling saving, searching, and loading project context directly from conversations.
README
Daruka MCP Server
Connect Claude directly to your Daruka project memory. Once installed, just tell Claude to save, search, or load your project context.
What it does
Claude gets four new tools it can call automatically:
| Tool | What it does |
|---|---|
daruka_list_projects |
Shows your Daruka projects so you can pick one |
daruka_save_memory |
Saves the current conversation + extracts decisions/bugs/TODOs |
daruka_get_context |
Loads your full Switch Context document into Claude |
daruka_search_memories |
Searches past memories by topic or keyword |
daruka_get_memories |
Lists your recent memory timeline |
Setup (5 minutes)
Step 1 — Get your Daruka token
- Go to daruka.vercel.app/settings
- Scroll to Chrome extension token
- Click Generate token
- Copy the token (starts with
ctxos_)
Step 2 — Create the MCP server folder
mkdir daruka-mcp
cd daruka-mcp
Copy these files into it:
daruka-mcp-server.tspackage.jsontsconfig.json
Then install and build:
npm install
npm run build
Step 3 — Add to Claude Desktop
Open your Claude Desktop config file:
- Mac:
~/Library/Application Support/Claude/claude_desktop_config.json - Windows:
%APPDATA%\Claude\claude_desktop_config.json
Add this (replace the path and token):
{
"mcpServers": {
"daruka": {
"command": "node",
"args": ["/absolute/path/to/daruka-mcp/build/daruka-mcp-server.js"],
"env": {
"DARUKA_TOKEN": "ctxos_your_token_here",
"DARUKA_API_URL": "https://daruka.vercel.app"
}
}
}
}
Important: Use the full absolute path — not ./ or ~/.
- Mac example:
/Users/yourname/daruka-mcp/build/daruka-mcp-server.js - Windows example:
C:\\Users\\yourname\\daruka-mcp\\build\\daruka-mcp-server.js
Step 4 — Restart Claude Desktop
Quit and reopen Claude Desktop. You should see a tools icon (🔧) in the chat interface.
How to use it
Just talk to Claude naturally:
"Save this conversation to my OutPitch project in Daruka"
"Load my ContextOS project context from Daruka"
"What did Daruka save about the Stripe webhook fix?"
"Show me my recent Daruka memories for this project"
"Save everything we discussed to Daruka — project is OutPitch"
Claude will automatically call the right Daruka tool.
Example conversation
You: Save this whole conversation to Daruka. Use my ContextOS project.
Claude: I'll save this to Daruka now.
[calls daruka_save_memory automatically]
✓ Saved to Daruka — "Stripe webhook debugging session"
Extracted: 2 decisions, 1 bugfix, 3 TODOs, 2 code patterns
Decisions: Use rawBody for constructEvent; server-side webhooks via Cloud Functions TODOs: Test retry logic with Stripe CLI; handle failed_payment event
View it at: https://daruka.vercel.app/project/your-project-id
Add it to Claude.ai (web) — coming soon
Claude.ai's web interface doesn't support MCP yet publicly, but Anthropic has announced MCP connectors for claude.ai. When that launches, Daruka will be submittable as an official connector. The same server code will work.
Troubleshooting
"DARUKA_TOKEN environment variable is required"
→ Make sure the token is in the env block of your claude_desktop_config.json
Claude doesn't show the tools icon → Check the config file is valid JSON (no trailing commas) → Make sure you used the full absolute path → Restart Claude Desktop completely (quit from taskbar/menu bar)
"Daruka API error 401" → Your token has expired or been revoked → Go to daruka.vercel.app/settings and generate a new one
Tools show up but save fails → Make sure you have a Gemini API key set in Daruka Settings → The extraction pipeline needs it to process the conversation
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。