AI coding agents are creating more documentation than ever β plans, TODOs, architecture notes, and implementation docs. The problem is that those docs don't always stay true, and the next agent may still use them as context. Docgrity finds duplicates, contradictions, and unanswered questions across your Markdown repos and Confluence, with the actual evidence behind every finding. Use it through VS Code, CLI, GitHub Actions, or the early-stage Confluence app. No finding without a receipt.
Hey Product Hunt! π
I built Docgrity after opening one of my own docs and realising I didn't believe a word of it anymore.
AI-assisted coding makes this problem a little stranger. Every agent session can leave behind another plan, TODO, architecture note, or temporary doc β and eventually those old docs become context for the next agent.
So I wanted a way to ask a simple question: **which parts of my documentation can I still trust?**
That's what Docgrity is trying to answer. It looks for duplicates, contradictions, and unanswered questions, but the thing I care about most is that every finding comes with the actual evidence behind it.
I'm still early, so I'd love to hear what you think. If you have a repo full of Markdown or a messy Confluence space, give it a try and tell me what it gets wrong β or what makes you say, "waitβ¦ that's still in there?" π