Dakini

Filter Criteria:

• Must contain .py files (not just __init__.py)
• Exclude common utility patterns: __pycache__, .pytest_cache, tests
• Minimum 2 Python files to qualify as a module

2. Module Analysis

For each discovered module, analyze:

Priority Detection:

# Critical: Contains domain models, core algorithms, or "main" logic
# High: Contains business logic, orchestration, or processing
# Medium: Contains utilities, factories, or helpers
# Low: Contains purely auxiliary functions

Indicators:

• Critical: Files with names like core.py, engine.py, repository.py, or contains Pydantic models
• High: Files with names like manager.py, processor.py, service.py
• Medium: Files with names like factory.py, builder.py, adapter.py
• Low: Files with names like utils.py, helpers.py, common.py

Description Generation: Read the module's __init__.py docstring or first .py file's module-level docstring to extract purpose.

Focus Areas Identification:

# Scan for common patterns:
- "repository" → repository_pattern
- "Protocol" class → clean_architecture
- "Builder" suffix → builder_pattern
- "Factory" suffix → factory_pattern
- "manager" → orchestration
- Statistical functions → algorithms

3. Configuration Generation

Template:

# Paper Generation Configuration
# Auto-generated by Dakini on {timestamp}