Environment and Setup Plan

Up: Environment · Curriculum: HVE Teaching Plan · Baseline: HVE Installation and Cohort Baseline

Desired Learner Experience

A learner receives one entry link, signs in with an approved account, opens a prepared workspace, completes a short health check, and starts the sample. Setup should consume little workshop time and give the instructor a clear recovery path.

Proposed target: most learners reach the health check within 15 minutes after preflight. This is a pilot target, not a measured guarantee. Do not promise one-click setup until the complete account and policy path has been tested.

Proposed Pilot Architecture

Use VS Code and GitHub Copilot with a prepared Python sample repository. Evaluate GitHub Codespaces as the primary hosted workspace and a local VS Code setup as fallback. Both should use the same source revision, runtime version, fixtures, and checks. Hosted access, organization policies, funding, and availability remain to be validated.

Begin with local synthetic-data exercises. Add Foundry and Fabric as separately provisioned modules once their learning outcomes require them; the HVE foundation should remain usable if a cloud service is unavailable.

Option	Benefit	Friction to test	Proposed use
Codespaces	Prepared workspace reduces machine differences	Account restrictions, network access, organization enablement, compute funding	First pilot candidate
Local VS Code	Useful for restricted organizations and repeat use	Runtime installation, corporate permissions, extension and proxy policies	Tested fallback
Learner’s client repository	Immediate work relevance	Confidentiality, inconsistent dependencies and approval delays	Later optional path

HVE Installation Choice

Microsoft’s installation guide supports the VS Code extension, Copilot CLI plugin, and selective clone adoption. The extension has release channels; the CLI registration tracks main. Canvases require the plugin and are not included in the extension. For the initial pilot, propose the extension without requiring canvases; verify its exact version and discovery of the required components. Reconsider the plugin only if a canvas becomes a learning requirement.

Avoid multiple installation sources for the same components until collisions have been tested. The published setup guide, selected baseline, and actual available components must agree.

Release Record

Complete before issuing learner instructions:

Field	Pilot value
Environment owner and support contact	TBD
Workspace image / devcontainer revision	TBD
Operating system and architecture	TBD
Python runtime	Prototype uses standard library; exact tested runtime to be recorded
Starter source revision	TBD; prototypes currently in vault
VS Code, Copilot, HVE versions	TBD
Model and relevant organization policies	TBD
Installation source and update policy	TBD
Learner distribution / instructor solution separation	TBD
Last clean-environment rehearsal and results	Not performed

Setup Journey

Before the Cohort

1. Program owner confirms learner count, funding, account ownership, and permitted tooling.
2. Learner receives prerequisites, access instructions, expected setup time, and a support route.
3. Learner verifies sign-in, Copilot access, workspace access, and ability to use the selected model. Record access failures without collecting credentials.
4. Instructor rehearses using a learner-equivalent account and a fresh workspace. Package only starter assets; keep solutions in an instructor-only location.
5. Freeze the tested source and tool baseline for delivery. Document how updates will be controlled and how to rebuild it.

Workshop Health Check

• Workspace opens and the expected source revision is visible.
• Python runs, sample input loads, and the sample’s documented baseline checks execute.
• Copilot can answer a bounded repository question with a relevant file reference.
• RPI Agent and the selected phase entry points are available.
• A learner can edit a scratch file, run a check, and reset to a clean starter copy.
• Instructor can distinguish an intentional failing acceptance check from an environment failure.

Recovery

Symptom	First action	Fallback
Sign-in or model access fails	Check approved account and organization policy with support	Pair with a working learner; record that independent tool use was not observed
Hosted workspace unavailable	Check organization access and connectivity	Tested local workspace
HVE entry point missing	Verify selected installation and reload as documented	Instructor-assisted artifact exercise; mark HVE execution incomplete
Code no longer runs after edits	Preserve learner work, compare diff	Fresh starter copy, then reapply the bounded change
Foundry/Fabric unavailable	Check module-specific provisioning	Continue the local HVE samples

Do not silently score a fallback demonstration as independent hands-on completion.

Evidence, Data, and Cleanup

Use synthetic fixtures only for the pilot. Keep task evidence locally during work; explicitly export the required plan, changes, review, and validation results for evaluation. Do not rely on a Git commit to capture ignored .copilot-tracking/ artifacts. Confirm submission destination, access, retention period, and deletion owner before collecting learner work.

For hosted workspaces, choose budgets and resource limits, idle shutdown, and deletion ownership before invitations. Cloud modules also need resource inventory and teardown verification. Do not put credentials in the repository, fixtures, prompts, or submitted evidence.

Planning Gates

• Approve the pilot host and fallback after testing both.
• Confirm access entitlements and per-learner cost ownership.
• Build and freeze a reproducible environment.
• Rehearse all samples from clean learner accounts and record failures.
• Publish the learner setup checklist and instructor recovery sheet.
• Measure setup time and revise the target after the pilot.

This document plans setup; it does not provision accounts, purchase seats, or create cloud resources.