Skip to content

ai-assisted-engineering.md

Latest commit

 

History

History
226 lines (153 loc) · 13.2 KB

ai-assisted-engineering.md

File metadata and controls

226 lines (153 loc) · 13.2 KB
Raw
title AI-Assisted Engineering
description Guide for using AI-powered tools like GitHub Copilot when working with the AI on Edge Flagship Accelerator
author Edge AI Team
ms.date 2026-05-15
ms.topic how-to
estimated_reading_time 5
keywords
ai-assisted engineering
github copilot
development workflows

AI-Assisted Engineering

This guide covers how to effectively use AI-powered tools, particularly GitHub Copilot, when working with the AI on Edge Flagship Accelerator.

Official Documentation

For comprehensive information about GitHub Copilot and VS Code integration, refer to the official documentation:

Project-Specific AI Resources

This repository includes specialized configurations and resources to enhance AI assistance:

Copilot Instructions

The repository includes comprehensive GitHub Copilot instructions in .github/copilot-instructions.md that provide:

  • Automatic Context Discovery: AI automatically finds and uses relevant project context
  • Convention Enforcement: Ensures all AI-generated code follows project standards
  • Component Understanding: Deep knowledge of the project's component and blueprint architecture
  • Markdown Standards: Automatic compliance with documentation formatting requirements

These instructions are automatically applied to every Copilot interaction, ensuring consistent, high-quality assistance.

Repository AI Guidance Files

The project contains specialized AI guidance files organized across different directories:

Core Guidance (/copilot/)

Comprehensive guidance files referenced by the main copilot instructions:

  • deploy.md - Deployment guidance and best practices
  • getting-started.md - Getting started guidance for new contributors
  • bicep/bicep.md - Edge AI Bicep deployment guidance
  • bicep/bicep-standards.md - Edge AI Bicep patterns and best practices
  • terraform/terraform.md - Edge AI Terraform deployment guidance
  • terraform/terraform-standards.md - Edge AI Terraform patterns and best practices

Note: Shared coding standards for Bash, Bicep, C#, commit messages, Markdown, Python scripting, Rust, and Terraform are provided by the hve-core VS Code extension and loaded automatically when installed.

Context Instructions (/.github/instructions/)

Instruction files designed to be attached to Copilot context using Add Context > Instructions:

File Name Context/Language Description
application.instructions.md Edge applications Edge application creation, import, and management guidance
build-documentation.instructions.md Build documentation Build and CI/CD documentation requirements
css.instructions.md Documentation CSS Modular CSS architecture and documentation site styling standards
javascript.instructions.md JavaScript Edge AI JavaScript guidance for backend, frontend, and utility code
rust-crate-registration.instructions.md Rust CI registration Rust crate registration requirements for CI test, coverage, and Codecov
tf-variable-consistency-manager.instructions.md Terraform governance Terraform variable validation and standardization workflow
wasm-build-deploy.instructions.md WASM operators WASM operator build, deploy, graph schema, and validation standards
wasm-operator-templates.instructions.md WASM operators Rust-based WASM operator templates
wasm-sdk-reference.instructions.md WASM operators WASM SDK reference patterns for Azure IoT Operations dataflow graph operators

Note: Shared task implementation, task research, task review, and learning coaching workflows are available through the hve-core VS Code extension.

Reusable Prompts (/.github/prompts/)

Prompt files for specific tasks that can be invoked using /prompt-name in Copilot chat:

Prompt Name Invocation Description Use Case
deploy.prompt.md /deploy Deployment workflows and best practices Infrastructure deployment assistance
getting-started.prompt.md /getting-started Project onboarding and initial setup guidance New contributor onboarding
edge-ai-project-planning.prompt.md /edge-ai-project-planning Edge AI project discovery and planning Scoping edge AI scenarios and solution capabilities
iotops-version-upgrade.prompt.md /iotops-version-upgrade Azure IoT Operations version upgrade process Updating IoT Ops components to latest versions
terraform-from-blueprint.prompt.md /terraform-from-blueprint Converting blueprints to Terraform Translating blueprint designs to infrastructure code
tf-variable-consistency-manager.prompt.md /tf-variable-consistency-manager Terraform variable consistency workflow Standardizing Terraform variables across components

Note: Additional prompts for ADR creation and prompt engineering are available through the hve-core VS Code extension.

Enhanced Custom Agents (/.github/agents/)

Advanced agent files with comprehensive tool access for specialized coaching and workflow assistance:

  • wasm-operator-builder.agent.md - Rust-based WebAssembly operator implementation for Azure IoT Operations dataflow graphs

Use the Edge AI Project Planning prompt for project discovery, scoping, and solution framing.

Note: Shared agents for ADR creation, task planning, task research, PR review, security planning, workback planning, implementation support, and prompt engineering are available through the hve-core VS Code extension.

Using Repository AI Resources

Applying Context Instructions

  1. Use Copilot Chat: Add Context > Instructions > Select the instruction file
  2. Add your specific context (files, folders, etc.)
  3. Provide your development prompt
  4. Instructions are automatically applied to ensure consistency with project standards

Invoking Reusable Prompts

  1. In VS Code, use Command Palette: Chat: Run Prompt and select desired prompt
  2. Or type /prompt-name directly in Copilot chat (e.g., /pull-request, /getting-started)
  3. Follow the guided workflow provided by the prompt

Using Enhanced Custom Agents

Custom agents provide specialized AI coaching with enhanced tool access, changing the system prompt in addition to the instructions:

  1. Reference Custom Agents: Use the agent drop-down in Copilot Chat to select a custom agent
  2. Choose Local Edge AI Resources: Select repository agents for WASM operator development or the project planning prompt for project scoping
  3. Use Shared HVE Core Agents: Use the hve-core extension for ADR creation, task planning, task research, PR review, security planning, workback planning, implementation support, and prompt engineering
  4. Enhanced Capabilities: Custom agents have comprehensive tool access for research, file editing, and system interaction

Task Planning and Implementation

  • Task Planner Custom Agent: Access advanced planning capabilities through the hve-core VS Code extension

    • Creates structured development plans with phases and tasks
    • Performs research to gather context for comprehensive planning
    • Generates documentation in ./.copilot-tracking/plans/ (excluded from git)

  • Task Implementor Agent: Access implementation workflows through the hve-core VS Code extension

    • Provides guidance for executing plans and tracking progress
    • Works with task planning outputs for coordinated development flow
    • Follows standardized workflows for consistent implementation practices
    • Uses .copilot-tracking/plans/, .copilot-tracking/details/, and .copilot-tracking/changes/ artifacts for implementation tracking

Learning AI Coaching Integration

Explore advanced AI-assisted engineering practices through our Learning Platform:

Interactive Learning Support

  • Task Check-offs: Mark progress and track learning automatically
  • Coaching Hints: Get contextual help when stuck on exercises
  • Smart Guidance: Personalized coaching based on your development patterns
  • Skill Assessment: AI-powered recommendations for your next learning steps

Getting Started with AI Coaching

  1. Launch Training Mode: Run npm run docs to access the learning platform
  2. Select Coaching Mode: Use the learning platform guidance to choose the appropriate coaching flow
  3. Start Learning: Say "I'm working on learning and want interactive coaching"
  4. Get Personalized Path: Take the skill assessment for customized kata recommendations

Learning coaching resources are pre-configured and ready to use in this repository. Shared advanced agents are available through the hve-core VS Code extension.

Essential Project Prompts

Pull Request Generation (/pull-request)

  • Generates comprehensive PR descriptions following project standards
  • Ensures proper documentation updates and review checklist completion
  • Options: includeMarkdown=true, branch=feat/branch-name

Task Planning

  • Task Planner: Available through the hve-core VS Code extension
  • Files stored in ./.copilot-tracking/ (excluded from git)
  • Works with the HVE Core Task Implementor agent for tracked implementation

Deployment Assistance (/deploy)

  • Provides deployment guidance and workflows specific to project blueprints
  • Infrastructure deployment assistance following project conventions

Architecture Decision Records

  • Guided ADR creation is available through the hve-core VS Code extension
  • Ensures proper documentation of architectural decisions

Project Structure Integration

The AI resources are designed to work with the project's specific structure:

Component Development

  • AI understands the decimal naming convention (e.g., 000-cloud, 010-security-identity)
  • Recognizes internal modules and their scoping rules
  • Follows deployment patterns from CI directories and blueprints

Blueprint Creation

  • AI can suggest component combinations based on existing blueprints
  • Understands output-to-input mapping between components
  • Follows blueprint documentation requirements

Framework-Specific Guidance

  • Terraform: Module organization, variable patterns, testing with Terratest
  • Bicep: Parameter definitions, module structure, Azure resource patterns
  • C#: Testing standards, project structure, dependency patterns

GitHub Copilot for Azure Extension

When using the Dev Container, the GitHub Copilot for Azure (Preview) extension provides:

  • Azure-specific agents: Use #azure... tags for Azure-specific assistance
  • Resource schema: #azureBicepGetResourceSchema for latest Bicep schemas
  • Best practices: #azureTerraformBestPractices for Terraform guidance
  • Documentation: #azureRetrieveMsLearnDocumentations for up-to-date Azure docs

Additional Resources

For general GitHub Copilot usage, refer to the official documentation.

🤖 Crafted with precision by ✨Copilot following brilliant human instruction, then carefully refined by our team of discerning human reviewers.