AI Automation

Put an AI Agent Inside GitHub Issues

October 14, 2026 Akshay Patel

If your tracker already lives in GitHub, this is the easiest one in the whole series to install. The issue and the code are already in the same place, so most of the plumbing the other recipes need simply is not required.

What it does

  • Add your analyze label to an issue, and the AI posts a breakdown grounded in the repository it is sitting in.
  • Comment back, and it answers in the thread and settles the scope.
  • Swap to your build label, and it opens a pull request.
  • No relay, no webhook, no extra token — it runs on native GitHub Actions triggers only.

What it will not do

It never relabels, never closes issues, and never merges its own pull request.

Start here if you are only trying one

Because there is no relay to deploy and no third-party token to mint, this recipe is the shortest path from curiosity to a working agent. Every run is a GitHub Actions run, which means the prompt, the full transcript and the result are uploaded as artifacts you can read afterwards — the whole thing is auditable in a tab you already have open.

Setup takes about five minutes

Clone the repo, run ./setup.sh, pick GitHub Issues, and answer a few questions — your IDs, your tokens, your conventions. The installer copies the recipe into your repository, sets your secrets, and prints whatever manual steps are left with your real values already filled in. No server to host, no database to run, nothing to pay for beyond the AI you already use.

Honest status: beta. It is built and smoke-tested, and it carries a live-fire checklist in its README that I am working through one tool at a time. It is the simplest recipe here, so it is also the best one to try first. If you hit an edge, open an issue — that is exactly what the beta label is asking for.

The recipe is MIT-licensed and open source: ai-automations. It is one of 35 — see the series overview for the full list and the rules every recipe follows. If your tool is not on the list, name it and I will build the automation for it, free.

FAQ

What do I have to deploy?

Nothing. No relay, no webhook, no extra token — it runs on native GitHub Actions triggers, which is why it is the easiest recipe in the series to install.

Will it relabel or close issues?

No. It never relabels, never closes, and never merges its own pull request.

Can I see what the agent actually did?

Yes. Every run is a GitHub Actions run, so the prompt, the full transcript and the result are uploaded as artifacts you can read afterwards.

More in this series: Monday.com · GitLab Issues · all 35 automations

Working on something similar?

If you are wiring AI into the tools your team already uses, I am happy to look at it with you. Thirty minutes, no deck — bring the workflow that costs your team the most time and we will work out whether it is automatable. Get in touch.