Documentation that keeps up with your code.
A GitHub App that documents every pull request automatically. Per-file READMEs, directory summaries, a master index, and live architecture diagrams — all committed back to your repo, reviewable in PR, never out of date.
500,000 tokens free. No credit card. Paid tiers from $10/month launching soon.
Every codebase ends up with a documentation folder nobody trusts. Last updated three years ago, written by someone who left, contradicted by the actual code. Manual documentation loses to delivery pressure every time.
Install DocBot on a repository. When a PR is opened, DocBot reads the diff, generates documentation for the changed files, and commits it to a _docs/ folder in your repo. Your PR review shows code changes and doc changes side by side. Merging the PR ships both.
Function signatures, purpose, structure, and dependencies. Code references stay accurate as the file changes.
Module overviews showing what each folder contains and how the pieces fit together.
A navigable map of the entire repository, regenerated as the codebase grows.
Mermaid diagrams generated from your code: system architecture, workflows, data models, component dependencies, API integrations. GitHub renders them natively in your repo.
Every package from package.json, requirements.txt, Cargo.toml and others, with licence data fetched from npm, PyPI, and crates.io. Generated as DEPENDENCIES.md for compliance audits and SBOMs.
Use .github/wai-docbot.yml to control docs paths, branch filters, and exclusions, or require approval with a token estimate before docs are generated on each PR.
From a pull request to finished docs: DocBot estimates the cost, generates the documentation on approval, and commits a structured _docs/ folder — per-file READMEs, directory indexes, and architecture diagrams. Click any screenshot to take a closer look.
DocBot runs on Azure OpenAI's GPT-5-mini — a deliberately cost-efficient model, far cheaper than the GPT-4 / Claude-class models most "AI docs" tools default to. That cost choice is intentional: documentation generation is a structure-and-accuracy task, not a creative-writing task, and mini holds up well for it.
The economics are what make the free tier genuinely usable rather than a 7-day trial.
| Tier | Price | Tokens / month | Status |
|---|---|---|---|
| Free | $0 | 500K lifetime + shared 250M pool | ✅ Live |
| Entry | $10/mo | 500K | Launching soon |
| Small | $50/mo | 3M | Launching soon |
| Medium | $150/mo | 10M | Launching soon |
| Enterprise | $500+/mo | 50M+ | Launching soon |
No per-seat fees on any tier. Add unlimited team members. Pay only for tokens used.
Free tier covers roughly 80–100 files of documentation lifetime, plus access to a shared 250M monthly pool for ongoing PR documentation. Small repositories run comfortably on free forever.
Whose READMEs visibly rot in public.
Who can't justify a dedicated technical writer.
Who need current documentation and dependency manifests for audits.
Who want better repo context for Copilot, Cursor, and Claude Code.
Generated docs become high-quality context for GitHub Copilot, Cursor, Claude Code, and other AI coding agents. When Cursor needs to answer "how does authentication work in this repo?", it reads _docs/ — and gets a current, accurate answer instead of stale README guesses. Up-to-date READMEs and architecture diagrams in your repo mean better suggestions, fewer hallucinations, and faster onboarding for both humans and AI tools.
Full setup guide: woden-ai.com/setup/docbot
500,000 free tokens, no credit card, no subscription. Paid tiers from $10/month launching soon.