Eject Phase 3: Standalone Operations Handoff

Parent: #206

Summary

Make it possible for the customer to actually run their infrastructure post-eject using the PTD CLI independently.

Approach

Hand over the PTD CLI + config (option 1). The customer runs ptd ensure themselves with control room fields stripped and standalone mode enabled.

Work Involved

• Strip control room fields from ptd.yaml during eject
• Document which ptd ensure steps to run and when
• Provide a standalone mode flag or config that skips control-room-dependent logic
• Pulumi state ownership runbook (state backend path, pulumi stack select setup)

Future Options (only if demand arises)

• Standalone Pulumi project — export Pulumi programs with hardcoded config, no PTD wrapper
• Terraform generation — convert resource inventory to Terraform import blocks + HCL

Open Questions

• Team Operator post-eject: frozen version at eject time, ongoing chart access, or customer builds their own?
• Observability replacement: standalone Grafana/Mimir/Loki stack, or just documentation?

Dependencies

• Phase 2 (control room severance) should be complete first

Start with the PTD CLI's ptd.yaml handling and the ptd ensure workflow described in the issue. Review how control-room fields, standalone mode, and Pulumi state setup are represented, then define the handoff steps and runbook. Done means a customer can run ptd ensure independently with the documented Pulumi stack selection and state backend setup.