# MCP Tools Directory
All MCP server capabilities are implemented as **tools** for maximum compatibility with AI agents.
## 🔧 **Tool Development Guidelines**
### **Required Tool Documentation Format**
**CRITICAL**: All tools MUST use this structured format for agent compatibility:
```python
def your_tool_function(
input_param: str,
optional_param: str = "default"
) -> Dict[str, Any]:
"""
TOOL_NAME=your_tool_function
DISPLAY_NAME=Human-Readable Tool Name
USECASE=When/why to use this tool (specific scenarios)
INSTRUCTIONS=Step-by-step usage guide for agents
INPUT_DESCRIPTION=Expected data format with examples
OUTPUT_DESCRIPTION=What format you'll receive back
EXAMPLES=your_tool_function("example_input", "optional_value")
PREREQUISITES=What to do first (workflow sequence)
RELATED_TOOLS=Other tools to use with this one
Traditional docstring for developers goes here...
"""
try:
# Input validation
if not input_param:
raise ValueError("input_param is required")
# Your business logic here
result = process_input(input_param, optional_param)
return {
"status": "success",
"operation": "your_operation",
"result": result,
"message": "Operation completed successfully"
}
except Exception as e:
return {
"status": "error",
"operation": "your_operation",
"error": str(e),
"message": "Operation failed"
}
```
### **Tool Registration**
Add your tool to `../mcp.py`:
```python
# Import your tool
from template_mcp_server.src.tools.your_tool import your_tool_function
# Register in _register_mcp_tools()
self.mcp.tool()(your_tool_function)
```
## 📋 **Current Tools**
- `multiply_tool.py` - Basic arithmetic operations
- `code_review_tool.py` - Generate code review prompts (converted from prompt)
- `redhat_logo_tool.py` - Asset retrieval (converted from resource)
## ✅ **Best Practices**
1. **Consistent Returns**: Always return `Dict[str, Any]` with `status` field
2. **Error Handling**: Wrap in try/catch, return structured errors
3. **Input Validation**: Validate all inputs before processing
4. **Logging**: Use `from template_mcp_server.utils.pylogger import get_python_logger`
5. **Testing**: Add tests to `../../tests/test_tools.py`
## 🎯 **Agent-Friendly Tips**
- Use **clear, action-oriented names** (`generate_report` not `report_generator`)
- Include **concrete examples** in EXAMPLES field
- Specify **prerequisites** for workflow guidance
- List **related tools** to help agents chain operations
- Keep **error messages** descriptive but concise
