Integrate Sentry with AI Agents

 To get Sentry set up for production error tracking:

1. Create a Sentry project
- Sign up at [sentry.io](https://sentry.io) (free tier covers most apps)
- Create an organization and a React project
- Note these three values:
  - SENTRY_ORG — your org slug (from the URL: `https://<org>.sentry.io`)
  - SENTRY_PROJECT — the project slug
  - SENTRY_AUTH_TOKEN — Settings → Auth Tokens (Organization Tokens). Just click "Create New  (needs `project:releases` and `org:read` scopes)

2. Add them to your AI developer Agent
Add these as Build Secrets in your workspace:
- Go to Workspace Settings → Build Secrets
- Add `SENTRY_AUTH_TOKEN`, `SENTRY_ORG`, and `SENTRY_PROJECT`
- These are injected during `bun run build`, so the source-map upload plugin can authenticate

3. About “automatically fixing errors”
Sentry does not auto-fix code by itself. What it can do:
- Capture errors with stack traces, release version, and user context
- Group similar errors and alert you
- The AI developer Agent Sentry MCP connector can read Sentry issues inside chat and help investigate root causes
- AI developer Agent can then propose and implement fixes based on that investigation

So the workflow is: error appears in Sentry → AI developer Agent connector surfaces it → you or a developer asks AI developer Agent to investigate → AI developer Agent proposes a fix → you approve and it lands in the codebase.