NVIDIA / NVIDIA/NemoClaw

design(onboard): define health-before-delete replacement

Open
#7,704 0 comments 0 reactions 1 assignee Claimed by @cv View on GitHub
area: architecture area: onboarding area: sandbox needs: design
Dominant language
TypeScript
Stars
22.5k
Forks
3.1k
Avg merge
1d 1h
Merged PRs (30d)
715

Description

## Ownership

@cv owns this issue.

Do not start, assign, or submit this work unless @cv transfers ownership in this issue.

## Problem

Generic replacement can delete the active sandbox before the replacement proves runtime health.

A prebuilt context proves buildability and input identity. It does not prove replacement health or atomic rollback.

## Required design

Define:

- the identity and naming model for a staged replacement;
- health checks that must pass before deletion or route change;
- the swap boundary for registry, gateway, policy, and forwards;
- rollback behavior after each swap effect;
- OpenShell constraints that prevent a true atomic operation;
- operator-visible recovery when rollback cannot restore service.

## Change boundary

This issue owns design and validation planning first.

Do not implement replacement behavior until a maintainer accepts the design.

Do not mix this work with the FSM, entrypoint, desired-state, or receipt stacks.

## Security

The design must preserve sandbox isolation, policy containment, credential boundaries, and route ownership during staging and swap.

## Completion

An accepted design defines observable behavior, failure recovery, compatibility, and E2E evidence for implementation.

Contributor guide

Open the contributing guide

Assessment

This issue has not been assessed yet.

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.