[NemoClaw][spark] Re-onboard with existing sandbox name silently overwrites instead of rejecting

[eko-liu]

Description

[Description]
When performing re-onboarding in NemoClaw, selecting an existing sandbox with the same name does not trigger a rejection or warning.
Instead of notifying the user that a sandbox with the same name already exists, the system silently overwrites the existing sandbox configuration.
This unexpected overwrite may lead to accidental data or configuration loss and inconsistencies during sandbox reinitialization.

[Environment]
Device: spark
Node.js: v20.20.1
npm: 10.8.2
Docker: Docker Engine 29.3.0
OpenShell CLI: 0.0.15
NemoClaw: v0.1.0 (cec1e4)
OpenClaw: 2026.3.11 (29dc654)
Note: Issue reproducible intermittently, primarily when re-onboarding via CLI after partial uninstall or cleanup.

[Steps to Reproduce]

1. Run onboarding to create a new sandbox environment, e.g. sandbox-a.
2. Once onboarding is complete, run re-onboard with the same sandbox name sandbox-a.
3. Observe that the system does not prompt any rejection or duplication warning.
4. After completion, sandbox-a is overwritten with the new configuration and previous data is lost.

[Expected Result]
When attempting to re-onboard using a sandbox name that already exists, the CLI should:

• Prompt the user with a confirmation or warning message.
• Optionally reject the request unless explicitly confirmed (e.g. via --force).

[Actual Result]
The CLI proceeds with the onboarding silently and overwrites the existing sandbox environment without any user confirmation or rejection message.

Steps to reproduce

No steps provided.

---

[NVB# 6029373]