MCP-IPFS
A Model Context Protocol server that enables Claude to interact directly with IPFS for content-addressed storage, peer-to-peer networking, and decentralized data management.
README
<p align="center"> <img src="https://raw.githubusercontent.com/ipfs/logo/master/ipfs-logo.svg" alt="IPFS Logo" width="120" /> </p>
<h1 align="center">MCP-IPFS</h1>
<p align="center"> <strong>A Model Context Protocol server for IPFS operations</strong> </p>
<p align="center"> Give Claude full control over the InterPlanetary File System </p>
<p align="center"> <a href="#-quick-start">Quick Start</a> • <a href="#-features">Features</a> • <a href="#-tools">Tools</a> • <a href="#%EF%B8%8F-configuration">Configuration</a> • <a href="#-architecture">Architecture</a> • <a href="#-contributing">Contributing</a> </p>
<p align="center"> <img src="https://img.shields.io/badge/license-MIT-blue.svg" alt="License: MIT" /> <img src="https://img.shields.io/badge/MCP-compatible-brightgreen.svg" alt="MCP Compatible" /> <img src="https://img.shields.io/badge/IPFS-Helia-blue.svg" alt="IPFS: Helia" /> <img src="https://img.shields.io/badge/TypeScript-5.0+-3178c6.svg" alt="TypeScript 5.0+" /> <img src="https://img.shields.io/badge/tools-34-orange.svg" alt="Tools: 34" /> </p>
What is MCP-IPFS?
MCP-IPFS is a Model Context Protocol (MCP) server that enables Claude and other AI assistants to interact directly with IPFS (InterPlanetary File System). It provides a comprehensive suite of tools for content-addressed storage, peer-to-peer networking, and decentralized data management.
Built on Helia (the modern JavaScript IPFS implementation), with a pluggable architecture ready for Kubo and remote pinning services.
Why MCP-IPFS?
| Without MCP-IPFS | With MCP-IPFS |
|---|---|
| Manual IPFS CLI commands | Claude handles IPFS operations directly |
| Copy-paste CIDs between tools | Seamless content-addressed workflows |
| No visibility into node status | Full observability of peers, bandwidth, DHT |
| Long operations block interaction | Background jobs with progress tracking |
| Single IPFS backend | Pluggable: Helia, Kubo, pinning services |
🚀 Quick Start
Prerequisites
- Node.js 18+
- Claude Code CLI or any MCP-compatible client
Installation
# Clone the repository
git clone https://github.com/EricGrill/mcp-ipfs.git
cd mcp-ipfs
# Install dependencies
npm install
# Build
npm run build
Add to Claude Code
Add to your ~/.claude/claude_desktop_config.json:
{
"mcpServers": {
"ipfs": {
"command": "node",
"args": ["/path/to/mcp-ipfs/dist/index.js"]
}
}
}
Verify Installation
You: What's the IPFS node status?
Claude: [Uses node_status tool]
Node is running with peer ID QmXxx...
Connected to 12 peers
Repository size: 45.2 MB
✨ Features
<table> <tr> <td width="50%">
📦 Content Operations
- Add & retrieve text, JSON, files, directories
- Pin management to prevent garbage collection
- IPNS publishing for mutable names
- MFS (Mutable File System) for familiar file operations
- DAG operations for advanced IPLD data structures
</td> <td width="50%">
🔭 Node Monitoring
- Node identity and health status
- Peer connections with latency and direction
- Bandwidth statistics total and per-protocol
- Repository stats and garbage collection
- DHT operations for peer/content discovery
</td> </tr> <tr> <td>
⚡ Background Jobs
- Async operations for large files and slow networks
- Progress tracking with real-time updates
- Job management - list, poll, cancel
- Auto-detection - small ops return immediately
</td> <td>
🔌 Pluggable Backends
- Helia (embedded) - zero external dependencies
- Kubo (planned) - connect to existing go-ipfs nodes
- Pinning Services (planned) - Pinata, Web3.Storage
</td> </tr> </table>
🛠 Tools
Content Tools
| Tool | Description |
|---|---|
ipfs_add |
Add text, JSON, or file to IPFS. Returns CID. |
ipfs_get |
Retrieve content by CID |
ipfs_pin |
Pin content to prevent garbage collection |
ipfs_unpin |
Unpin content |
ipfs_pins_list |
List all pinned CIDs |
ipfs_add_directory |
Add entire directory, returns root CID |
ipfs_ls |
List contents of directory CID |
ipfs_get_file |
Get specific file from directory by path |
IPNS Tools
| Tool | Description |
|---|---|
ipns_publish |
Publish CID to IPNS name |
ipns_resolve |
Resolve IPNS name to current CID |
ipns_keys_list |
List available IPNS keys |
MFS Tools (Mutable File System)
| Tool | Description |
|---|---|
mfs_write |
Write content to mutable path |
mfs_read |
Read content from MFS path |
mfs_ls |
List directory contents |
mfs_mkdir |
Create directory |
mfs_rm |
Remove file or directory |
mfs_mv |
Move/rename file or directory |
mfs_flush |
Flush MFS to get root CID |
DAG Tools
| Tool | Description |
|---|---|
dag_put |
Store IPLD data, returns CID |
dag_get |
Retrieve IPLD node, optionally traverse path |
Node Tools
| Tool | Description |
|---|---|
node_id |
Get peer ID, public key, versions |
node_status |
Quick health check |
node_version |
Detailed version info |
peers_list |
Connected peers with details |
peers_count |
Simple peer count |
bandwidth_stats |
Network bandwidth metrics |
swarm_connect |
Connect to peer by multiaddr |
swarm_disconnect |
Disconnect from peer |
repo_stats |
Repository statistics |
repo_gc |
Trigger garbage collection |
DHT Tools
| Tool | Description |
|---|---|
dht_find_peer |
Find addresses for peer ID via DHT |
dht_find_providers |
Find peers providing a CID (configurable limit) |
dht_provide |
Announce content availability to the network |
Job Tools
| Tool | Description |
|---|---|
job_status |
Get status of background job |
jobs_list |
List all jobs |
job_cancel |
Cancel running job |
⚙️ Configuration
Configuration File
Create mcp-ipfs.config.json in the project root:
{
"backend": "helia",
"persistence": {
"enabled": true,
"path": "./ipfs-data"
},
"helia": {
"libp2p": {
"addresses": ["/ip4/0.0.0.0/tcp/4002"],
"bootstrap": true
}
},
"jobs": {
"maxConcurrent": 3,
"defaultTimeout": 300000
}
}
Environment Variables
Environment variables override config file settings:
| Variable | Description | Default |
|---|---|---|
MCP_IPFS_BACKEND |
Backend to use | helia |
MCP_IPFS_PERSISTENCE_ENABLED |
Enable persistent storage | true |
MCP_IPFS_DATA_PATH |
Data directory path | ./ipfs-data |
MCP_IPFS_BOOTSTRAP |
Connect to bootstrap nodes | true |
MCP_IPFS_JOBS_MAX_CONCURRENT |
Max concurrent jobs | 3 |
MCP_IPFS_JOBS_TIMEOUT |
Default job timeout (ms) | 300000 |
Persistence Modes
Persistent (default):
- Content survives server restarts
- Uses LevelDB blockstore
- Recommended for production
In-memory:
- Content lost on restart
- Faster for testing
- Set
persistence.enabled: false
🏗 Architecture
┌─────────────────────────────────────────────────┐
│ Claude │
└─────────────────────┬───────────────────────────┘
│ MCP Protocol (JSON-RPC)
┌─────────────────────▼───────────────────────────┐
│ MCP-IPFS Server │
│ │
│ ┌───────────┐ ┌───────────┐ ┌─────────────┐ │
│ │ Content │ │ Node │ │ Job │ │
│ │ Tools │ │ Tools │ │ Manager │ │
│ └─────┬─────┘ └─────┬─────┘ └──────┬──────┘ │
│ │ │ │ │
│ └──────────────┼───────────────┘ │
│ │ │
│ ┌───────────▼───────────┐ │
│ │ Backend Layer │ │
│ │ (IPFSBackend) │ │
│ └───────────┬───────────┘ │
└───────────────────────┼─────────────────────────┘
│
┌─────────────┼─────────────┐
┌─────▼────┐ ┌─────▼────┐ ┌─────▼────┐
│ Helia │ │ Kubo │ │ Pinning │
│(embedded)│ │ API │ │ Services │
└──────────┘ └──────────┘ └──────────┘
✓ planned planned
Key Components
| Component | Responsibility |
|---|---|
| Content Tools | IPFS add/get/pin, IPNS, MFS, DAG operations |
| Node Tools | Status, peers, bandwidth, repo, DHT, config |
| Job Manager | Queue long operations, track progress, enable polling |
| Backend Layer | Abstract IPFS implementation details |
Backend Interface
interface IPFSBackend {
// Lifecycle
start(): Promise<void>;
stop(): Promise<void>;
// Content
add(content: Uint8Array, options?: AddOptions): Promise<CID>;
get(cid: CID): Promise<Uint8Array>;
pin(cid: CID): Promise<void>;
unpin(cid: CID): Promise<void>;
// Node info
id(): Promise<PeerInfo>;
peers(): Promise<PeerConnection[]>;
// ... etc
}
📁 Project Structure
mcp-ipfs/
├── src/
│ ├── index.ts # Entry point
│ ├── server.ts # MCP server setup
│ ├── tools/
│ │ ├── content.ts # Content operations
│ │ ├── node.ts # Node monitoring
│ │ └── jobs.ts # Job management
│ ├── backends/
│ │ ├── interface.ts # Backend abstraction
│ │ └── helia.ts # Helia implementation
│ ├── jobs/
│ │ └── manager.ts # Job queue
│ └── config.ts # Configuration
├── test/
│ ├── content.test.ts
│ ├── node.test.ts
│ └── jobs.test.ts
├── docs/
│ └── plans/ # Design documents
├── package.json
├── tsconfig.json
└── mcp-ipfs.config.json
🧪 Development
# Install dependencies
npm install
# Run in development mode
npm run dev
# Run tests
npm test
# Build for production
npm run build
# Lint
npm run lint
🤝 Contributing
Contributions are welcome! Please read our contributing guidelines before submitting PRs.
- Fork the repository
- Create a feature branch (
git checkout -b feature/amazing-feature) - Commit your changes (
git commit -m 'Add amazing feature') - Push to the branch (
git push origin feature/amazing-feature) - Open a Pull Request
📚 Related Resources
Learn More
Community
📜 License
MIT License - see LICENSE for details.
🙏 Acknowledgments
- Protocol Labs for IPFS and Helia
- Anthropic for Claude and MCP
- The agents-skills-plugins marketplace for inspiration
<p align="center"> Built with ❤️ for the decentralized web </p>
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。