Configurable LLM Client Settings

[hmshuv]

Configurable LLM Client Settings

Summary

Introduces a configuration system using pydantic-settings to allow users to customize LLM client behavior (retry logic and default models) via environment variables, replacing hardcoded values.

Changes

New Module:

• Added strategicwm/_src/config.py with Settings class for configuration management

Updated Modules:

• Modified strategicwm/_src/client_lib.py to use config.settings.RETRY_* instead of hardcoded values

Dependencies:

• Added pydantic-settings to pyproject.toml

Testing:

• Added strategicwm/_src/config_test.py to verify configuration overrides

Motivation

Hardcoded retry attempts (tries=10) and delays (delay=10) made the library inflexible for different network conditions or model requirements. This change allows users to tune parameters without modifying code.

Related Issue

Closes #11

Type of Change

• Bug fix
• New feature
• Breaking change
• Documentation update

Usage Example

# Set custom retry behavior
export SWM_RETRY_COUNT=5
export SWM_RETRY_DELAY=15
export SWM_MODEL_NAME=gpt-4

# Run your code - settings are automatically loaded
python your_script.py

Checklist

• Code follows project style guidelines
• Self-reviewed code
• Added tests proving feature works
• All tests pass locally
• Documentation updated

---

Ready for review! Let me know if you have any questions.