CodexMaker
Problem
Documentation is the most universally acknowledged gap in software projects. The reason it stays undocumented is not laziness — it is that documentation is a separate authoring process disconnected from the code it describes. The code changes. The documentation does not update. The drift between code and docs becomes the documentation itself: unreliable.
Approach
CodexMaker automates documentation generation by extracting information directly from the codebase: comments, function definitions, type signatures, module structures, and file organization. It produces a static documentation website that reflects the actual current state of the code, not a manually maintained approximation.
The output is navigable, searchable, and structured by the same hierarchy as the codebase. When the code changes, regenerating documentation produces an updated site. The documentation cannot drift because it is derived, not authored.
Implementation
The tool parses repository file trees, extracts JSDoc/TSDoc comments, function signatures, export structures, and module relationships. It generates a static site with sidebar navigation mirroring the repository structure, syntax-highlighted code blocks, and cross-linking between referenced modules.
Available as an npm package and deployable via GitHub Pages or any static hosting. The generation process runs in CI/CD pipelines, producing updated documentation on every merge to main.