Help centre · Getting started

How does Diavi know which technology it is working on?

The stack detected then confirmed for each repository, skill: labels, and issues that touch several technologies.

Each repository has its stack, which Diavi suggests by reading its files and you confirm. An issue then follows that stack's conventions, unless a skill:<name> label on the issue says otherwise; with neither, Diavi uses the account's technology.

How do I confirm a repository's stack?

Preferences → Pipeline → Execution → Stack of each repository: pick the repository, click Detect, then tick and confirm. Diavi only reads the file tree and a few manifests, never your code, and saves nothing until you have confirmed. The first stack ticked is the main one.

In what order does Diavi decide?

1. the stack chosen at launch, if you forced it;
2. a skill:<name> label on the issue;
3. the repository's confirmed stack;
4. a keyword recognised in the labels;
5. the account's technology.

What about an issue that touches several technologies?

It can carry up to three. Each file is written with the rules of its own stack, and checking is done stack by stack. Beyond three, Diavi refuses and asks you to split the issue: cutting a rule in half would be worse than not sending it.

What if the stack is wrong?

The run produces code written to another technology's conventions. It breaks nothing — the merge request stays a proposal — but the review will flag it. Fix the repository's stack or the label, and run again.

Where can I see the available technologies?

The list is published on the product page, and it is generated from the real configuration: it cannot announce a technology that does not exist.

And my in-house library, my own conventions?

They are not about the technology but about your context, and you can write them yourself. See My contexts.