Context

Build the configuration loading system that reads YAML files and validates them against Pydantic models.

Acceptance Criteria

• Load company config from YAML file(s)
• Validate against Pydantic models with clear error messages
• Support layered config: base defaults + user overrides
• Environment variable substitution in config values (${ENV_VAR})
• Config file discovery (current dir, ~/.ai-company/, etc.)
• Helpful error messages pointing to exact YAML location of validation failures
• Unit tests for valid configs, invalid configs, missing fields, env var substitution

Tech Decisions Needed

• YAML library: PyYAML vs ruamel.yaml (round-trip, comments)
• Config merging strategy (deep merge vs shallow)

Dependencies

• Core entity models issue (M1)

Design Spec Reference

Sections 4.3, 5.4, 9.2, 10.4