Ask AI to fix a bug, right inside GitHub

A small thing that changes how you interact with AI

Right now, if you want an AI to help with something in your project — fixing a bug, reviewing a document, updating a task — you probably open a separate chat window, paste in all the context, wait for an answer, and then go back and do the work yourself.

OpenTag skips that whole dance.

You're in a GitHub issue or a Slack thread. You type something like @agent fix the broken checkout button — and the agent picks it up, does the work on its own copy of the project, and posts back the result (including a ready-to-approve fix) right in that same thread. No switching windows. No losing track of what was said.

What makes it feel trustworthy: nothing leaves your own machine. Your code, your passwords, your project files — they stay local. That matters when you're working with clients or handling anything sensitive.

It also works with several different AI models — Claude, Codex, Gemini — so you're not forced to bet on one forever.

This is early software (version 0.1.0, launched last week), so it'll have rough edges. But the idea is sound, and teams that try it now will have a head start on a way of working that feels genuinely different.

Something to sit with: How much time does your team spend switching between tools to get a single thing done? That friction adds up faster than we notice.

Words worth knowing

Agent — An AI that doesn't just answer questions, but actually takes actions: writing code, creating files, making changes on your behalf.

Pull Request (PR) — A formal proposal in GitHub to add changes to a project. Think of it like a tracked edit in Google Docs, but for code.

Open-source — Software whose inner workings are publicly visible. Anyone can inspect it, which tends to build trust — and means the community can fix problems quickly.

Local-first — The software runs on your own computer, not on someone else's server. Your data doesn't travel anywhere you haven't approved.