Remediations
A diagnosis that ends at “you should probably” is half a product
Decim proposes the change, links it to the evidence that justified it, and improves it as review comes back. Nothing is ever merged automatically.
Not every fix is a pull request
A file replay is not a code change, and dressing it up as one helps nobody.
Pull requests
A branch, a diff, and a rationale citing the evidence that motivated it. Draft by default, never merged automatically.
Data changes
The mapping row that was missing — applied where the pipeline actually reads it, and to the repository copy as well.
Configuration
A scheduler window moved, a retry policy narrowed. No diff, because there is no code to change.
Operational actions
Replay these files, in this order. Decim proposes it and explicitly cannot perform it.
They get better, and the record says why
A revision with no stated cause is indistinguishable from a rewrite at random.
v1
The initial proposal
Derived from the diagnosis, with the evidence that justified it attached.
v2
After a review comment
A reviewer objects, and the revision records what they said and what changed because of it.
v3
After a failing check
CI catches a test asserting the old behaviour. The next revision states that as its cause.
Get started
Bring us a pipeline that broke last week
The fastest way to evaluate this is a real incident you already know the answer to. If Decim gets it wrong, that is a far more useful demo than one where it doesn't.