Tool-Use Repository Development Guidelines

Creating a New Script

To add a new script to the tool-use repository, you need to modify several files:

1. Create a new script file in src/tool_use/scripts/:

Example: src/tool_use/scripts/my_script.py from ..utils.ai_service import AIService from ..config_manager import config_manager def main(args): config_values = config_manager.get_tool_config("script_name") # Your script implementation pass if name == "main": try: main() except KeyboardInterrupt: console.print("\n[yellow]Operation cancelled by user.[/yellow]") sys.exit(0) except Exception as e: console.print(f"\n[red]An unexpected error occurred: {e}[/red]") sys.exit(1)

2. Add dependencies to src/tool_use/scripts/_script_dependencies.py:

SCRIPT_DEPENDENCIES = {

... existing dependencies ...

"my-script": ["required-package1", "required-package2"] }

3. Add configuration in src/tool_use/utils/config_wizard.py: SCRIPT_INFO = {

... existing scripts ...

"my-script": { "name": "My Script", "description": "What your script does", "config_keys": [ { "key": "my_config_key", "prompt": "Enter configuration value:", "description": "What this config value does", "required": True } ] } }

4. Update src/tool_use/cli.py:

• Add to all_scripts dictionary
• Add to script_modules mapping

5. Add script to README.md

Using AIService

The AIService utility (src/tool_use/utils/ai_service.py) provides a standardized way to interact with AI models. Example usage from prioritize.py:

from ..utils.ai_service import AIService

Initialize the service

ai = AIService()

Simple completion

response = ai.query("Your prompt here")

#Structured output

class Example(BaseModel): example: str = Field(description="example")

class StructuredOutput(BaseModel): """Example description for the structured output, keys can be any pydantic supported field""" field1: str = Field(description="First field for the output") field2: List[Example] = Field(description="List of examples")

Structured outputs only supported by openai

ai_service = AIService(service_type="openai") response = ai_service.query_structured(prompt, StructuredOutput, system_prompt)

field1, field2 = response

Script Best Practices

1. Use config_manager for settings:

from ..config_manager import config_manager config_values = config_manager.get_tool_config("script_name")

2. Handle arguments properly: def main(args): print("Usage: tool-use my-script [args]") return

3. Provide clear error messages and user feedback

4. Follow existing patterns for consistency:

• Use relative imports
• Handle dependencies through SCRIPT_DEPENDENCIES
• Provide configuration through config_wizard
• Document usage in script's docstring

Error Handling

Make sure to handle errors gracefully and provide clear error messages to the user.

Formatting

• Use Rich when possible to display print statements in a visually appealing way

Common Patterns

1. AI Integration:

   • Use AIService for AI interactions
   • Structure prompts clearly
   • Handle API errors gracefully

2. Configuration:

   • Use config_wizard for user settings
   • Validate config values
   • Provide clear configuration prompts

3. CLI Integration:

   • Follow existing argument patterns
   • Provide help text
   • Handle invalid inputs gracefully