obsidian-ai-curator
Smart note consolidation for Obsidian using Claude AI. Automatically identifies and merges scattered fragments into structured knowledge via MCP. Like Tetris for your thoughts - pieces fall into place.
Install / Use
claude mcp add nwant -- npx -y github:nwant/obsidian-ai-curatorIf the server publishes to npm under a different name, use that package instead — check the repo README.
MCP Server
Model Context Protocol server
Quality Score
Category
AutomationSupported Platforms
Skill content
View source on GitHubObsidian AI Curator
The knowledge fragmentation problem: Your insights are scattered across hundreds of notes. The solution: AI-powered consolidation that actively merges related fragments into coherent knowledge. How it works: Claude AI analyzes your Obsidian vault via MCP and intelligently combines scattered thoughts.
What It Does
Gives Claude direct access to your Obsidian vault. Ask Claude to:
- 🔍 "Search for notes about machine learning"
- 📝 "Create a new project note with proper formatting"
- 🏷️ "Find all notes tagged #active that need review"
- 📁 "Move completed projects to archive"
- 🔄 "Create a git checkpoint before making changes"
- 🔗 "Rename this file and update all links"
- 📅 "Add a task to today's daily note"
Quick Start
1. Install MCP Server
git clone https://github.com/nwant/obsidian-ai-curator.git
cd obsidian-ai-curator
npm install
2. Configure Your Vault Path
cp config/config.minimal.json config/config.json
Edit config/config.json and set your vault path:
{
"vaultPath": "/path/to/your/obsidian/vault"
}
3. Add to Claude Desktop
Edit your Claude Desktop config:
- macOS:
~/Library/Application Support/Claude/claude_desktop_config.json - Windows:
%APPDATA%\Claude\claude_desktop_config.json - Linux:
~/.config/Claude/claude_desktop_config.json
{
"mcpServers": {
"obsidian-vault": {
"command": "node",
"args": ["/absolute/path/to/obsidian-ai-curator/src/mcp-server.js"]
}
}
}
Important: Use the full absolute path, not relative paths or ~
4. Restart Claude Desktop
Completely quit and restart Claude Desktop to load the MCP server.
5. Verify Installation
In Claude, type:
What MCP tools do you have access to?
You should see tools like vault_scan, write_note, search_content, etc.
See Quick Start Guide for detailed instructions and optional plugin setup.
Features
- Search & Query: Natural language search, metadata queries, Dataview integration
- Smart Writing: Auto-formats links, validates tags, manages frontmatter
- File Management: Rename/move files with automatic link updates
- Tag Management: Analyze usage, validate tags, rename globally
- Project Playbooks: Customizable structures for different project types
- Git Integration: Checkpoint and rollback capabilities
- Daily Notes: Quick capture and task management
- 🚀 Automated Workflow: Claude Desktop → GitHub Issues → Claude Code → Pull Requests
See Examples for detailed use cases.
🚀 NEW: Automated Claude Code Integration
Turn errors into fixes and ideas into features - automatically!
This project now includes TWO ways to automate development with Claude Code:
Option 1: Local Execution (Recommended)
Run Claude Code on your machine - no GitHub Actions needed!
# Quick setup
bash scripts/setup-local-claude.sh
# Requirements:
# 1. Install Claude Code CLI from https://claude.ai/code
# 2. Install GitHub CLI: brew install gh
# 3. Authenticate both tools
Benefits:
- ✅ No GitHub Actions costs
- ✅ Runs immediately on your machine
- ✅ Full control and visibility
- ✅ Easy debugging
See Local Claude Code Guide for setup.
Option 2: GitHub Actions Automation
Use GitHub Actions to run Claude Code in the cloud
# Setup GitHub Actions workflow
bash scripts/setup-automation.sh
# Requires Claude Code OAuth token in GitHub Secrets
Benefits:
- ✅ Runs in cloud, not on your machine
- ✅ Triggered automatically by issues
- ✅ Works even when you're offline
See GitHub Actions Workflow Guide for setup.
Documentation
- Quick Start Guide - Get running in 5 minutes
- Local Claude Code - Run Claude Code locally (recommended)
- GitHub Actions Workflow - Cloud-based automation
- Troubleshooting - Common issues and solutions
- Configuration Guide - All configuration options
- MCP Tools Reference - Complete tool API documentation
- Examples - Common use cases and workflows
- Project Playbooks - Creating custom project playbooks
- Obsidian Plugin Guide - Plugin features and setup
- Formatting Rules - Important Obsidian formatting guidelines
Project Structure
This repository contains:
- MCP Server (
src/) - Core functionality for Claude integration - Obsidian Plugin (
obsidian-ai-curator-plugin/) - Enhanced API performance - Documentation (
docs/) - Comprehensive guides
Requirements
- Node.js 18+
- Obsidian (for vault)
- Claude Desktop or Claude Code
- Git (optional, for version control features)
Common Issues
"Claude doesn't see the MCP tools"
- Did you restart Claude completely after editing the config?
- Is the path to
mcp-server.jsabsolute in your Claude config?
"Permission denied" errors
- Check that your vault path in
config.jsonis correct - Ensure you have read/write permissions to your vault
"Cannot find module" errors
- Run
npm installin the project directory - Make sure you're using Node.js 18 or higher
See Troubleshooting Guide for more help.
Testing
This project uses Jest as the test runner with a comprehensive test suite:
# Run all tests
npm test
# Run specific test suites
npm run test:unit # Unit tests (parallel execution)
npm run test:integration # Integration tests (sequential execution)
npm run test:coverage # With coverage report
npm run test:watch # Watch mode for development
npm run test:verbose # Detailed output
# Additional testing tools
npm run test:benchmark # Performance benchmarks
npm run test:harness # Verify test infrastructure
Jest Benefits:
- ✅ Reliable integration tests - No vault conflicts
- ✅ Fast unit tests - 4x parallel execution
- ✅ Enhanced error messages - Better debugging
- ✅ Modern tooling - Watch mode, coverage, verbose output
See test documentation for details on writing and running tests.
Contributing
Contributions welcome! Please read our contributing guidelines before submitting PRs.
License
MIT
Related Skills
Agent-Reach
84.4kGive your AI agent eyes to see the entire internet. Read & search Twitter, Reddit, YouTube, GitHub, Bilibili, XiaoHongShu — one CLI, zero API fees.
headroom
73.4kCompress tool outputs, logs, files, and RAG chunks before they reach the LLM. 20% fewer tokens for coding agents, 60-95% fewer tokens for JSON, same answers. Library, proxy, MCP server.
ruflo
73.0k🌊 The original agent harness. Deploy intelligent multi-player swarms, coordinate autonomous workflows, and build conversational AI systems. Features adaptive memory, self-learning intelligence, federation, vector RAG integration, and native Claude Code / Codex / Hermes and many more Integrated
CowAgent
47.1kOpen-source super AI assistant & Agent Harness. Plans tasks, runs tools and skills, self-evolves with memory and knowledge. Multi-agent, multi-model, multi-channel. Lightweight, extensible, one-line install.
