Frequently asked questions

Does my code ever leave my Mac?

No. With on-device triage, everything runs locally using Apple's Foundation Models, so your diffs, PR contents, and metadata never leave your machine. The only exception is if you bring your own AI provider, in which case the PRs you triage go to the endpoint you chose, and nowhere else.

What do I need to run it?

MergeMole runs on any Mac with macOS 15 or later and a GitHub account. The free, on-device triage uses Apple's Foundation Models, which need macOS 26 (Tahoe) on an Apple Silicon Mac (M1 or later). On anything older you can still bring your own provider or run with no AI at all for a clean, organized PR queue.

Is it safe to give MergeMole a GitHub token?

Yes. You connect by pasting a personal access token (classic) with read access, which MergeMole verifies before saving and stores in your macOS Keychain, never in plain settings. It only ever reads. MergeMole never approves, comments on, merges, or writes anything to your repositories.

Which pull requests show up?

The ones that involve you, gathered across every repo and org you can access: PRs requesting your review, assigned to you, that you opened, that mention you, or that you have already reviewed. They merge into one deduped list, and tabs let you filter by how each one relates to you.

How does MergeMole decide priority and effort?

Prioritycomes from reading each PR's full context, its description, diff, CI status, comments, labels, and more, to flag what's urgent or high priority. Effort comes from the line count and PR content, sized from small (10 minutes or less) to extra large (an afternoon). Every verdict also carries one clause of why, so it is never a black box.

Do I have to use AI, and can I use my own?

AI is optional and flexible. Run the free on-device triage, bring your own provider (OpenAI, Anthropic, Ollama, LM Studio, or any OpenAI-compatible endpoint), or turn AI off entirely for a clean, organized list of the PRs that need you.

How much does it cost?

The on-device triage and the no-AI mode are completely free, with no limits. If you bring your own AI provider, you simply cover your own usage with that provider.

Still have a question? Reach out at awln.dev.