FAQ

Common questions.

Why does ocra report so few findings?

By design: reviewers are told what not to flag, findings are validated and fact-checked against the code, and findings outside a task's files are dropped. An empty report on a correct change is a good outcome.

How much does a review cost?

It depends on the models and the change. In the evaluation runs (Gemini on Vertex AI, list prices) a review cost $1.57 on average (median $1.15, from $0.07 to $3.98); see Measured quality. ocra review --plan estimates without calling a model, --max-cost-usd caps a run, and every run prints tokens and cost.

Which models can I use?

Any model OpenCode supports, or your own OpenAI-compatible endpoint, written as provider/model; see Model providers. A failback chain per tier keeps an overloaded model from failing the review.

Does it work with GitHub pull requests?

Yes: ocra review --pr <number>, or the GitHub Action, which posts inline comments and a summary and re-reviews only what changed. See GitHub pull requests.

Does it work with GitLab?

Yes, GitLab.com and self-managed: ocra review --mr <iid> posts threads and a summary note, from CI or the command line. See GitLab merge requests.

Why doesn't the model give line numbers?

Models often get them wrong, so they quote the code and ocra resolves the quote to lines. See How it works.

Where are results stored?

In .ocra/sessions/<id>/ of the reviewed repository; see Sessions.

Edit on GitHub

On this page