GitMir Compute takes the work your team accepts on a project, does it from the same Intelligence your people and agents use, and commits the result to your repository.
A person decides what gets done. GitMir Compute does it with the product context already in hand, and a person checks what came back.
Work is written down on the project’s board with what must be true when it is done. An Admin of the project accepts it — nothing is carried out before a person does.
For each accepted piece GitMir Compute asks the project’s Intelligence for the part of the product it concerns, the same way your own agent would.
The work is done on GitMir’s infrastructure and committed to your repository — straight to the branch, or as pull requests if you switch them on.
The piece waits in Verify for a person. Its commits show on Changes, and the model updates after them like after any other push.
Nothing to set up twice: GitMir Compute works on the project’s repositories, people, access and model. The only new permission is write access to those repositories through its GitHub App.
To do · In progress · Verify · Done
Commits straight to the branch by default · pull requests if you switch them on
The project’s Intelligence — what the parts are and what depends on them
Its own GitHub App, with write access to the repositories you give it
Deploy. It commits; your pipeline does the rest
$99 per project / month, however many repositories the project holds. Two projects are 2 × $99.
Next to your GitMir Intelligence plan, on any of them
People on the project do not add to the price
Model updates after its commits use your team’s credits, like any other push
It runs to the end of the paid period, and the work stays on the board
GitMir Compute works from a project’s Intelligence. Connect your repository first, then add GitMir Compute to the project from the cabinet.