Our team has less time to document things, but we need better notes on what projects are for and why they were built. Is there a tool or workflow that can turn a rough brain dump into useful docs? I'd review the result before sharing it.
What AI tools work well for writing semi-technical docs?
Asked by
overqualified-for-tetris
Answers
Try giving Claude Code a specific documentation template and asking it to inspect the relevant codebase. Specify the audience and how technical the result should be, then have it write a draft you can copy into your team's documentation system.
Code can show what a system does, but it won't necessarily explain why the team made a decision. Add your own notes for that part and check the draft against someone who knows the project.
Reply to this answer
For the brain-dump part, keep the workflow simple. Give an AI assistant your rough notes and ask it to organize them around purpose, users, key decisions, and anything still unknown. Tell it not to fill gaps with guesses.
Then check the draft against the notes and flag questions for someone who knows the project. A tidy explanation is no use if it confidently invents the reason something exists.
Reply to this answer
Another thing to explore is reusable open source skills for coding assistants. They may help standardize a documentation workflow across projects, though I'd inspect what a skill does before installing it and still review everything it produces.
Reply to this answer
Add your answer
Answers are reviewed before publication.
9c-delta:
Codex is another option for gathering context from a codebase. The resulting draft will still need editing and fact-checking.