Skip to main content

​
Build Powerful AI Tools with Simple JSON Schemas

Model Context Interface (MCI) transforms how you create AI agent tools. Define collections of tools in a single JSON file, execute them anywhere, and share them effortlessly. No complex servers, no vendor lock-in – just pure simplicity.

🚀 Get Started

Jump into the quickstart guide and build your first MCI tool in under 5 minutes with Python. Use it with any framework you like: ADK, LangChain, CrewAI, etc.

💻 Schema reference

Explore details about JSON schema - understand what capabilities does the MCI have and how it can contribute to your projects

​
Why Choose MCI?

Simple & Lightweight

Define tools declaratively in simple JSON files. No complex server setup, no vendor-specific APIs – just clean, readable schemas that anyone can understand and review.

Universal Compatibility

Works with every programming language and framework. Whether you’re using Python, Node.js, or any other language, MCI includes operations which are available in any programming language

Secure by Design

Easy to Review & Audit Your entire toolset fits in a single JSON file that’s easy to review – even by an LLM. No more black-box servers or third-party code accessing your sensitive data.

Multiple Execution Types

HTTP • CLI • File • Text: Wrap REST APIs, command-line tools, file operations, and text templates – all from the same unified interface. Perfect for integrating diverse systems and workflows.

Built-in Authentication

API Key • Bearer Token • Basic Auth • OAuth2 Comprehensive authentication support built right into the schema. Securely connect to any service without writing custom authentication logic.

Advanced Templating

Dynamic Values • Conditionals • Loops Powerful template engine with environment variable substitution, conditional logic, and iteration for complex, dynamic tool execution.
MCI makes AI tool development accessible to everyone by using JSON schemas that work in every programming language. It’s a lightweight alternative to complex systems like MCP, living right in your project repo.

​
The Vision: Simplifying AI Development

MCP is a great tool that allows to create servers and tools with complex logic, but in practice, it turned out that a lot of MCP servers are just API or CLI tool wrappers. And prompts? Have you ever used “prompts” feature of MCP? And it is the entire server in your infrastructure. So, sometimes MCP can be quite an overhead. MCI changes everything by bringing the simplicity and universality of JSON to AI tool development, acting more like a supplement to MCP than alternative. You know, HTTP request, terminal command & reading file are basic features that every programming language supports. And sometimes (pretty often), it is what the AI needs as a tool, but framed, described, standardized, reusable, shareable and secure. That’s what the MCI is!

​
How MCI Makes AI Development Better

1

Universal Standard

One Schema, Every LanguageJSON works everywhere. Your MCI tools work in Python today, Node.js tomorrow, and any future language your team adopts. No rewrites, no vendor lock-in.
2

Instant Shareability

Share your entire AI toolset by copying a single JSON file. Teams can collaborate on tools like they collaborate on code – with version control, reviews, and easy distribution.
3

LLM-Friendly Design

The simple JSON format means LLMs can help you build tools. For example, Share API documentation with an AI, and it can generate whole MCI schema in minutes – not hours or days.
4

Maximum Flexibility

One .mci.json file for your whole project, separate files per agent, one file per API, or mix tools from different authors. It doesn’t matter – it’s not 10 servers to initialize, it’s just 10 files in your repo.
5

Security & Trust

Transparent & AuditableEvery tool is defined in plain JSON that humans and AI can easily review. No hidden server code, no mysterious third-party services – just transparent, auditable tool definitions.

​
Quick Example

See how simple it is to create and use AI tools with MCI: 
{
  "tools": [
    {
      "name": "greet_user",
      "description": "Generate a personalized greeting message",
      "inputSchema": {
        "type": "object",
        "properties": {
          "username": {
            "type": "string",
            "description": "user's name"
          }
        },
        "required": ["username"]
      },
      "execution": {
        "type": "text",
        "text": "Hello, {{props.username}}! Welcome to MCI."
      }
    }
  ]
}
That’s it! You’ve just defined and executed your first MCI tool. Scale this to hundreds of tools across your entire AI application.

​
Coming Soon

The Python adapter is ready and battle-tested with 92%+ test coverage. More language adapters are on the way!

Node.js Adapter

In Development Full TypeScript support with the same simple API. Perfect for JavaScript and Node.js applications.

Go Adapter

Planned High-performance Go implementation for system-level tools and microservices.

PHP Adapter

Planned Bringing MCI to the PHP ecosystem for web applications and CMS integrations.

​
Sponsors

No Sponsors Yet – But You Can Change That!This project is built and maintained by individual developers with passion for simplifying AI development. We’re not backed by tech companies or funded in any way. Any help will be appreciated!
Want to support MCI’s development?
  • 🐛 Report bugs and suggest features
  • 💻 Contribute code, documentation, or examples
  • 📢 Spread the word and share the project
  • 💝 Become a sponsor to support ongoing development
We offer different sponsorship tiers with exclusive perks:
  • 🚀 Visibility in our channels
  • 🎯 Priority support for your use cases
  • 📚 Custom implementations and examples
  • 🏆 Recognition in our documentation and releases
Contact us: revaz@usemci.dev

​
Common Use Cases

MCI excels in scenarios where you need simple, reliable tool integration without the complexity of full server deployments.
Transform any REST API into an AI-friendly toolPerfect for wrapping third-party APIs like weather services, payment processors, or internal microservices. MCI handles authentication, request formatting, and response processing automatically.
{
  "name": "get_weather",
  "execution": {
    "type": "http",
    "method": "GET",
    "url": "https://api.weather.com/current",
    "auth": { "type": "apiKey", "in": "header", "name": "X-API-Key" }
  }
}
Automate infrastructure tasks with CLI tool integrationWrap command-line tools like Docker, Git, AWS CLI, or custom scripts. Perfect for deployment automation, monitoring, and system management tasks.
{
  "name": "deploy_service",
  "execution": {
    "type": "cli",
    "command": "docker",
    "args": ["deploy", "--stack", "{{props.service_name}}"],
    "cwd": "{{env.DEPLOY_PATH}}"
  }
}
Create data pipelines and processing workflowsChain together data transformation tools, database queries, and analytics scripts. Ideal for ETL processes, report generation, and data analysis workflows.
Build comprehensive tool libraries for AI agentsCreate specialized toolsets for different agent roles - customer service bots, development assistants, or data analysis agents. Easy to share and maintain across projects.
Orchestrate complex multi-step processesCombine different execution types to create sophisticated workflows. Perfect for marketing, content publishing, or business process automation.
Use File execution to process complex Markdown or XML prompts
{
  "name": "support_assistant_instructions",
  "execution": {
    "type": "file",
    "path": "./instructions.md",
    "enableTemplating": true
  }
}

Get Started Now with Python

Jump into the quickstart guide and build your first MCI tool in minutes

Join the Community

Connect with other developers and share your MCI tools
⌘I