NVIDIA / NVIDIA/NemoClaw

Assess Relay and Switchyard integration requirements for other agent harnesses

Open
#8,892 0 comments 0 reactions 1 assignee Claimed by @cjagwani View on GitHub
area: architecture area: routing
Dominant language
TypeScript
Stars
22.5k
Forks
3.1k
Avg merge
1d 1h
Merged PRs (30d)
715

Description

Parent Epic: #7937

## Outcome

Document what is reusable for other NemoClaw agent harnesses and what each harness must implement, without expanding Hermes delivery into unsupported runtimes.

## Scope

- Define the harness contract for native Relay lifecycle, model-call interception, configuration, session/feedback identity, routing metadata, credentials, readiness, and teardown.
- Compare OpenClaw and Deep Agents with the Hermes contract.
- Identify shared NemoClaw/OpenShell provider and policy work versus harness-specific adapters.
- Recommend separately approved follow-up issues; do not implement or claim support here.

## Acceptance

- [ ] The reusable Relay/Switchyard platform boundary is documented.
- [ ] Each assessed harness has a concrete gap list and owner recommendation.
- [ ] No Switchyard algorithm or provider-security logic is duplicated per harness.
- [ ] Any implementation work is routed through a separately accepted issue with its own E2E criteria.

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.