Python, JavaScript, and TypeScript at launch. Go, Rust, and Java are planned next. DocFlow uses AST analysis to help documentation follow code changes, but generated output still needs review.

DocFlow is not designed to store full repository source code. It may store repository metadata, configuration, generated documentation, and processing history, and it processes the code context needed to generate docs. Relevant context may be sent to DeepSeek, the current production AI provider.

DocFlow's current production configuration uses DeepSeek as the AI provider, with deepseek-chat as the default model. DocFlow first extracts code structure with AST analysis, then sends the limited context needed for a documentation draft to the configured AI endpoint. If the production provider or model changes, this disclosure will be updated.

Yes. DocFlow supports custom README templates, custom changelog formats, and configurable inline comment styles. You define the template once, and DocFlow applies it to every PR.

DocFlow posts its changes as a PR comment first. You review and edit before anything lands on main. You can also configure DocFlow to create a separate branch with its changes, giving you full control.

Free for public repos. Private repos are billed per PR processed. No monthly minimums. If you exceed your plan limits, DocFlow stops processing and notifies you — no surprise charges.

Not yet. DocFlow is currently a managed GitHub App. Self-hosting is on the roadmap for 2025.