mci-py 1.1.1

Python adapter for MCI

mci-py

MCI Python Adapter - A lightweight, Python adapter for the Model Context Interface (MCI), enabling AI agents to execute tools defined in JSON or YAML schemas.

The Model Context Interface (MCI) is an open-source, platform-agnostic system for defining and executing AI agent tools through standardized schemas.

Using the basic features that are supported in every programming language, MCI makes it easier to define collections of AI Tools, filter, execute and maintain. Making it a strong alternative or supplement of MCP, which lives right in your project repo and fits in single JSON or YAML file. (Check example.mci.json or example.mci.yaml & example.py)

The mci-py Python adapter allows you to load tool definitions from JSON or YAML files and execute them with full control over authentication, templating, and error handling.

---

Features

• 🚀 Simple API - Load and execute tools with just a few lines of Python code
• 📝 Schema-Based - Define tools declaratively and statically in JSON or YAML files
• 🔄 Multiple Execution Types - Support for HTTP, CLI, File, and Text execution
• ✔️ Easy to build - Share MCI Schema reference and documentation of any REST API or CLI application to LLM to build your favorite tools in minute
• 🔐 Built-in Authentication - API Key, Bearer Token, Basic Auth, and OAuth2 support
• 🔁 Easy to share - All you need to move your toolset to another project, or share it to the world is a single JSON or YAML file.
• 🎯 Template Engine - Dynamic value substitution for environment variables and properties, as well as advanced templating directives like "@if", "@foreach", etc. to support complex prompting.
• 🔒 Secure by design - All you need is adapter and schema file, which is easy to review, even by an LLM, compared to the whole servers. No more third-party code or servers accessing your data.
• 💔 Flexible separation - You can have one ".mci.json" file for whole project and filter tools out, or you can have 1 file per agent to include whole toolset of agent, or you can have 1 file per API wrapped, you can even have 10 files from different authors and use only 1 tool from each - you can do anything: it doesn't matter, it's not 10 MCP servers to initialize, it's just 10 files in your repo 🤷
• 🎨 Type Safe - Full type hints and Pydantic validation
• 🧪 Well Tested - 92%+ test coverage with comprehensive test suite
• ✅ Schema Validation Mode - Validate schemas without loading toolsets or MCP servers, perfect for CI/CD and development tools

Planned

• Switch template engine - Now mci-py have built-in basic template engine supporting variables, @if, @for & @foreach blocks. We are planning to add implement Jinja for more robust templating options and update JSON schema to define template engine
• Include - Add @include("path/to/file.md") directive to simplify reusing the prompt parts

---

Usage example

from mcipy import MCIClient

# Initialize with your JSON schema file
client = MCIClient(
    schema_file_path="my-tools.mci.json",
    env_vars={
        "API_KEY": "your-secret-key",
    }
)

# Or with YAML schema file
client = MCIClient(
    schema_file_path="my-tools.mci.yaml",
    env_vars={
        "API_KEY": "your-secret-key",
    }
)

# Get all tool objects
tools = client.tools()

# Execute the tool with properties
result = client.execute(
    tool_name="greet_user",
    properties={"username": "Alice"}
)

# Validating mode: Validate schema without env vars or side effects
# Perfect for CI/CD pipelines and development tools
validator = MCIClient(
    schema_file_path="my-tools.mci.json",
    validating=True  # No env vars needed!
)
print(f"Schema is valid with {len(validator.list_tools())} tools")

---

Documentation

• API Reference - Comprehensive API documentation
• Quickstart Guide - Get started quickly with examples
• Schema Reference - Complete schema documentation (JSON & YAML)

---

Examples

Explore the examples directory for comprehensive demonstrations:

• HTTP Examples - JSON | YAML - HTTP API calls with authentication
• CLI Example - Command-line tool execution
• File Example - File reading with templating
• Text Examples - JSON | YAML - Text template generation
• Mixed Examples - JSON | YAML - Combined execution types
• Python Usage Example - Complete Python integration example

---

Support

Need help or have questions?

• 📖 Check the Documentation section above
• 🐛 Open an issue for bug reports
• 💬 Start a discussion for questions and ideas
• 📧 Contact the maintainer: [e-mail]

---

Contribution

We welcome contributions! Here's how you can help:

1. Fork the repository
2. Create a feature branch (git checkout -b feature/amazing-feature)
3. Make your changes following our coding standards
4. Run tests and linting (make test && make lint)
5. Commit your changes (git commit -m 'Add amazing feature')
6. Push to the branch (git push origin feature/amazing-feature)
7. Open a Pull Request

Development Setup

To quickly set up your development environment, run:

./setup_env.sh

This will install uv, Python, and all project dependencies automatically.

Test Coverage

Run make coverage for coverage report

Project Docs

For how to install uv and Python, see installation.md.

For development workflows, see development.md.

For instructions on publishing to PyPI, see publishing.md.

---

Donation

This project is not backed or funded in any way. It was started by an individual developer and is maintained by the community. If you find this project useful, you can help in several ways:

• ⭐ Star the repository to show your support
• 🐛 Report bugs and suggest features
• 💻 Contribute code, documentation, or examples
• 📢 Spread the word and share the project
• 💝 Consider a donation to support ongoing development

Any kind of help is greatly appreciated! 🙏

---

Credits

• Author: MaestroError (Revaz Ghambarashvili)
• Contributors: Thank you to all the amazing contributors who have helped improve this project!
• Template: This project was built from simple-modern-uv