Doc Gen

Doc Gen creates comprehensive documentation by analyzing your source code and coordinating three specialized AI agents to produce structured, accurate docs that reflect your actual codebase.

Three-stage generation process

Doc Gen uses a pipeline approach where each stage handles a specific aspect of documentation creation:

Stage Agent Responsibility
Outline outline-planner Scans codebase structure and creates documentation hierarchy
Write content-writer Generates detailed content with code examples and API references
Polish polish-reviewer Reviews and refines the final output for clarity and consistency

The DocumentGenerationWorkflow orchestrates these agents, ensuring each focuses on its domain while producing a cohesive final document.

Chunked processing for large codebases

When documenting complex projects, Doc Gen breaks the work into manageable chunks to avoid overwhelming the AI agents:

This approach ensures that even extensive codebases get thorough documentation without hitting token limits or generating superficial coverage.

Output structure

Every Doc Gen run produces a structured markdown document with four consistent sections:

When to use Doc Gen

Run Doc Gen when you need comprehensive documentation that accurately reflects your current codebase structure. Unlike manually written docs that drift from the code, Doc Gen reads the actual source files and generates content based on what's really there — function signatures, module exports, class hierarchies, and import relationships.