Note: configuration

Context

Attune AI's configuration system provides unified settings management across all agents and workflows, with support for multiple file formats and environment variable overrides.

Configuration architecture

The configuration system centers on the UnifiedAgentConfig class, which normalizes settings for all agent types. The ConfigLoader class handles file discovery, loading, and saving, while maintaining backward compatibility with legacy configuration formats.

Configuration files are discovered automatically using a search path hierarchy:

  1. ./attune.config.json (project-local)
  2. ~/.attune/config.json (user-specific)
  3. ~/.config/attune/config.json (XDG standard)

Environment variable compatibility

The system maintains compatibility with legacy EMPATHY_ environment variables while transitioning to the ATTUNE_ prefix. The get_attune_env() function checks for ATTUNE_ variables first, then falls back to EMPATHY_ equivalents.

Agent specialization

While UnifiedAgentConfig provides the base configuration model, specialized configurations like BookProductionConfig extend it with agent-specific settings. The for_book_production() method creates these specialized configurations while preserving the unified structure.

Workflow integration

The WorkflowConfig class manages agent workflow execution, supporting different modes through the WorkflowMode enum. This allows workflows to run in development, production, or testing configurations without code changes.

Source files

Configuration management spans multiple modules in src/attune/config/, including agent configurations, environment compatibility layers, and the core configuration loader.

Tags: config, settings