🚀 Robust and Dev friendly MCP framework
This will be oss very soon, working on docs + other misc stuff, if you want to contribute, send me an email.
Using uv (recommended):
uv pip install axiom-mcp-
Clone the repository:
git clone https://github.com/axiomml/axiom-mcp.git cd axiom-mcp -
Install uv if you haven't already:
curl -LsSf https://astral.sh/uv/install.sh | sh -
Create and activate a virtual environment with uv:
uv venv --python 3.12 source .venv/bin/activate # On Unix/Linux # or .venv\Scripts\activate # On Windows
-
Install development dependencies:
uv sync --frozen --extra dev
Tools in Axiom MCP are defined as classes that inherit from the Tool base class. Here's how to define a tool:
from axiom_mcp.tools.base import Tool, ToolMetadata, ToolValidation
# Define input schema for tool validation
number_input_schema = {
"type": "object",
"properties": {
"a": {"type": "number", "description": "First number"},
"b": {"type": "number", "description": "Second number"},
},
"required": ["a", "b"],
}
class AddTool(Tool):
"""Tool for adding two numbers."""
metadata = ToolMetadata(
name="add",
description="Add two numbers together",
validation=ToolValidation(input_schema=number_input_schema),
author="MathServer",
version="1.0.0",
)
async def execute(self, args: Dict[str, Any]) -> Dict[str, Any]:
a, b = args["a"], args["b"]
result = a + b
return {
"type": "text",
"content": {"operation": "addition", "a": a, "b": b, "result": result},
}Axiom MCP provides comprehensive logging and metrics capabilities for tools:
Each tool has access to a context with built-in logging methods:
class MyTool(Tool):
async def execute(self, args: Dict[str, Any]) -> Dict[str, Any]:
# Access logging through tool context
self.context.debug("Debug level message")
self.context.info("Processing started")
self.context.warning("Warning message")
self.context.error("Error occurred")
# Your tool logic here
return resultThe tool manager automatically tracks:
- Total calls
- Successful calls
- Failed calls
- Average execution time
- Last used timestamp
Metrics are stored in the logs/tools.log file by default and include:
- Tool name
- Operation status
- Timestamp
- Execution time
- Error messages (if any)
- Cache Control:
# Disable caching for specific execution
tool_context = ToolContext(cache_enabled=False)
tool = MyTool(context=tool_context)- Execution Timeout:
# Set custom timeout
tool_context = ToolContext(timeout=30.0) # 30 seconds- Dry Run Mode:
# Enable dry run for testing
tool_context = ToolContext(dry_run=True)Resources are lightweight endpoints that can be defined using decorators:
from pathlib import Path
from axiom_mcp import AxiomMCP
mcp = AxiomMCP("MyServer", port=8888)
# Simple string resource
@mcp.resource("greeting://{name}")
def say_hello(name: str) -> str:
return f"Hello, {name}!"
# Resource returning a list of files
@mcp.resource("dir://desktop")
def list_files() -> list[str]:
desktop = Path.home() / "Documents"
return [str(f) for f in desktop.iterdir()]axiom-mcp dev server.py # This runs in dev mode
axiom-mcp run server.py # This runs in release/prod mode# Run tests
uv run pytest
# Run tests with coverage
uv run pytest tests
# Update dependencies
uv add <dep>
# Sync your environment
uv sync- Fork the repository
- Create a new branch for your feature
- Make your changes
- Run the tests:
uv run pytest
- Submit a pull request
GNU General Public License v3 (GPLv3)