March 14, 2026
Repo or it didn’t happen
It's time to move your docs into a repo – especially because of AI
Some say it’s old news, others fear AI snooping, one team loves auto‑updated docs
TLDR: The post urges teams to keep documentation with code, saying AI tools make updates easier and enforce accuracy. Commenters brawl: some say this is old news, others reject AI as a justification or worry about privacy, while a few boast their AI assistant already updates code and docs together.
A new post says: move your docs into the repo (the code folder) because AI (artificial intelligence) now makes keeping them fresh easy. The comments instantly split into camps. Veterans like prepend roll in with a shrug: “We’ve been doing this forever.” Purists like jaredcwhite fire back that “because of AI” isn’t a real reason to change how projects run. Privacy hawks such as jmclnx say version control is great—but “give AI access to your docs, no thanks.”
On team “heck yes,” themanmaran flexes a real setup: one big repo and an AI helper that can add a new API option and update the docs in one go. petcat drags off‑site wikis for going stale, arguing that docs next to code keep releases and write‑ups moving together. The post imagines docs becoming the product spec—think documentation‑driven development—and the thread turns it into a meme: “Let the AI intern fix the README at 3 a.m.” The mood? Three‑way drama: “old news,” “don’t justify it with AI,” and “AI already makes this awesome.” The one thing everyone agrees on: scattered, stale docs are chaos; keeping them in the repo forces reality checks and fewer nasty surprises.
Key Points
- •The article argues documentation should live in the code repository to leverage version control, proximity to code, and integrated review processes.
- •It lists tools (Sphinx autodoc, JSDoc, Javadoc, Docusaurus) that generate API docs from code, reducing manual duplication across external platforms.
- •Testing code examples within docs (e.g., Python doctest in CI) helps keep documentation accurate and executable.
- •AI agents increase markdown content and “rules” files; much of this content overlaps with documentation and could be consolidated as repo docs.
- •The author foresees a shift to reviewing specs, harnesses, and guidelines over LLM-generated code, making high-quality, human-reviewed docs critical.