Every release, your changes are reapplied intelligently. When they ship the same feature, autofork stops applying yours.
Runs on the ChatGPT or Claude plan you already pay for.
Each new release, your changes go back on top. When one no longer fits their new code, an agent rebuilds it from your words.
always show context usage
PATCH.md Always show the thread's context usage in the composer. Build it on whatever T3 Code shows for context at the time.Goes back on as is.
They redid the composer. The agent rebuilt yours into it.
They added a usage meter in a menu. Yours uses it and stays in sight.
An example on T3 Code. If they ship exactly what you asked for, yours pauses and you decide.
My T3 Code fork took 156 commits from T3 Code in its first two days, and I never rebased it. One change broke on day one and the agent rebuilt it.
Your fork lives in your GitHub account and updates itself in its own GitHub Actions, on the ChatGPT or Claude plan you already pay for. Free on public forks.
Rather run it yourself? The engine is open source. Use it with your own agent.
npx skills add AVGVSTVS96/autoforki have an agent-maintained herdr fork that tracks upstream and applies/rewrites patches–one of which adds support for pane id labels–and cuts patched nightlies.
just a fun excuse to test out automating repo maintenance with PATCH.md files
automatic repo maintenance/patching is a cool technique i’ve been using to make forking and modifying existing projects more approachable
bots track upstream and maintain the repo automatically
i think it would be good to package this so anyone can set up their own auto-maintained fork
i have an idea:
take open source repos that rarely merge/review contributor PRs and set up an auto maintenance bot to keep up with upstream while copying and merging good contributor PRs the maintainers never merge
auto patching/repo-maintenance is game-changing. i can fork and customize anything, the agents will autonomously track upstream and re-implement my changes