nemoclaw-maintainer-validate-launchable

Validates the staging Brev Launchable deployment, runtime identity, and inference end to end.

22.3k|3.1k|Updated Mar 15, 2026
One-click install
npx skills add https://github.com/NVIDIA/NemoClaw --skill nemoclaw-maintainer-validate-launchable
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: nemoclaw-maintainer-validate-launchable
Source: https://github.com/NVIDIA/NemoClaw/tree/main/.agents/skills/nemoclaw-maintainer-validate-launchable
Command: npx skills add https://github.com/NVIDIA/NemoClaw --skill nemoclaw-maintainer-validate-launchable

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Maintainers need a rigorous, advisory manual validation of the user-facing staging Brev Launchable deployment that automated E2E tests cannot fully cover, including the web deployment flow, image identity, runtime receipts, and inference checks.

Core Features & Use Cases

  • Web Journey Validation: Walks the Brev Launchable deployment flow in an authenticated browser, records instance type and price, and requires explicit approval before creating a billable instance.
  • Image and Runtime Identity Verification: Compares the deployed GCE instance image and /etc/nemoclaw/provision.json receipt against the launchable-e2e.json workflow artifact for the selected commit SHA.
  • Credential-Safe Inference Testing: Runs hosted and sandbox inference checks only with a short-lived NVIDIA_INFERENCE_API_KEY delivered through a supported secret mechanism, with rotation or revocation after the run.
  • Use Case: A maintainer hands Codex a deployed Brev environment URL and asks for validation; the Skill verifies access, image identity, the baked full E2E user journey, and produces a structured advisory report classifying the result as complete pass, failed, partially blocked, or not run.

Quick Start

Use the nemoclaw-maintainer-validate-launchable skill to validate the staging Brev Launchable deployment at the provided environment URL and report the result.

Frequently Asked Questions about nemoclaw-maintainer-validate-launchable

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I validate a Brev Launchable deployment manually?

Open the accepted Launchable URL in an authenticated browser, confirm the page loads and shows a deployment action, record the instance type and price, obtain explicit approval, then deploy one instance and verify the environment reaches a successful state with a usable access method.

How do I verify the deployed Brev instance runs the correct NemoClaw image?

Read the GCE instance image metadata and require it to equal the boot.bootImage URI recorded in the launchable-e2e.json artifact from the selected automated workflow job. Also check /etc/nemoclaw/provision.json and confirm the baked source checkout has no tracked or untracked changes.

Can I validate a Launchable without an NVIDIA inference API key?

Yes, but inference checks are skipped. Report deployment, access, image, and runtime results separately and classify hosted inference, sandbox inference, and the overall result as partially blocked due to unavailable inference credentials. Never ask the user to paste a key into chat.

What happens if the launchable-e2e.json artifact is missing or malformed?

Stop validation immediately. Classify the evidence as failed when GitHub is available and the job or artifact can be inspected, or partially blocked when GitHub or the artifact service is unavailable and no performed check failed.

Does this validation replace the automated Launchable E2E tests?

No. The report is advisory manual validation and must not be used as automated E2E evidence. It complements the automated Staging Brev Launchable workflow by covering the user-facing web journey and runtime identity checks.