What the analysis delivers
- A summary of the problem and the confidence of the analysis.
- The root cause, explained from the code.
- Where it is: repository, file and line, with a direct link to the line on GitHub or Azure Repos.
- A suggested fix and the code change (patch).
- How to validate the fix.
How the AI finds the right code
It uses the real context of the report: the failing route, the screen the user was on, the error messages and, if Sentry is connected, the server-side error. For UI bugs it locates the page through the front-end routes and focuses on what the report is about.
Your key, your control
- You choose the AI provider (OpenAI or Anthropic) and use your company’s own key.
- Turn each provider on or off and choose the model for the analysis.
- Code access is read-only and limited to the branch you choose.
- The analysis prompt is editable.
Where it works
Repositories on GitHub and Azure Repos. The comment goes to the Jira Software card, the Jira Service Management request or the Azure DevOps work item.
Frequently asked questions
Does the AI change my code?
No. Capto only reads the chosen branch with a read-only token and writes the analysis as a comment on the ticket. The suggested fix is for your team to review and apply.
Which GitHub token do I need?
A fine-grained token with access to the project repositories and the Contents: Read-only permission.
Which AI is used?
The one you configure: OpenAI or Anthropic, with your company’s key.