ops(observability): configure and verify Sentry in staging and production

Status

The Sentry client/code integration already exists. This issue tracks operational configuration and verification, not a normal product implementation PR.

Current repository state already includes:

• Sentry client initialization and error boundary/capture helpers
• source-map upload script using sentry-cli
• CI upload steps for staging and production
• graceful no-op/skip behavior when Sentry configuration is intentionally absent

As of this audit, gh variable list shows no repository variables and gh secret list does not include SENTRY_AUTH_TOKEN, SENTRY_ORG, or SENTRY_PROJECT, so this issue remains open.

Rollout

1. Create/confirm the Sentry organization and comapeo-cloud-app project.
2. Configure repository/environment settings expected by the existing workflows:
   • repository variable VITE_SENTRY_DSN
   • secret SENTRY_AUTH_TOKEN
   • secret SENTRY_ORG
   • secret SENTRY_PROJECT
3. Verify the staging deployment first.
4. Verify staging source-map upload succeeds.
5. Send a controlled staging/test error and confirm the event is received with a correctly symbolicated stack trace.
6. Only after staging verification, enable/verify the production path.
7. Verify production source-map upload succeeds.
8. Confirm at least one normal production event is observed when production naturally emits an event; do not intentionally crash production solely to satisfy this issue.
9. Update the deployment runbook with verification and rollback/disable steps.

Acceptance criteria

• Sentry organization/project created and project DSN obtained
• VITE_SENTRY_DSN configured in the intended GitHub repository/environment variable scope
• SENTRY_AUTH_TOKEN configured in the intended secret scope with only the permissions required for source-map/release upload
• SENTRY_ORG configured
• SENTRY_PROJECT configured
• Staging deploy completes with Sentry enabled
• Staging source-map upload succeeds
• A controlled staging/test error is received by Sentry
• The staging test event stack trace is correctly symbolicated to application source
• Production is enabled/verified only after staging validation
• Production source-map upload succeeds
• At least one normal production event is observed and symbolicated
• Absence of Sentry credentials/configuration does not break preview or other unprivileged CI paths that are intentionally allowed to run without Sentry
• Deployment runbook documents how to verify Sentry health and how to disable/roll back the integration safely

Guardrail

Do not intentionally crash production solely to produce an acceptance-test event. Use staging/test for controlled error injection and normal production traffic for production verification.

Start by inspecting the existing Sentry client initialization, source-map upload script, and staging/production CI steps described in the issue. Configure the listed repository variables and secrets, verify staging before production, confirm symbolicated events and credential-free paths, then update the deployment runbook with safe verification and rollback steps.