localmTUTS
FollowFollowSubscribe
Short Course · 9 lessons + 1 interview questions · ~95 mins

Context Engineering for GitHub Copilot

Master the seven customization surfaces — instructions, prompts, agents, skills, MCP, hooks, and plugins — that transform GitHub Copilot from a generic assistant into a domain expert for your project.

Context EngineeringGitHub Copilot.githubCustom InstructionsMCPAI-Assisted SDLC
BeginnerModerateExpert
What You'll Learn

The .github folder, end to end

How custom instructions activate via glob patterns, how prompt files create reusable slash commands, how agents define specialized personas — the complete directory anatomy, explained through a real-world project.

Seven surfaces, one context architecture

Build each customization layer individually — instructions, prompts, agents, skills, MCP servers, hooks, and plugins — then combine them into a composable three-axis context system that activates automatically.

Full SDLC integration

Apply context engineering across the entire software development lifecycle: plan with prompts, code with instructions, review with agents, and automate with hooks — from first commit to production CI or CD.

VS Code and CLI throughout

Every demo runs in VS Code with GitHub Copilot Chat and the Copilot CLI. No cloud accounts or paid APIs required — you build a real .github setup that works on your local machine from lesson one.

About This Course

GitHub Copilot is a powerful code generation tool. But out of the box, it does not know your project's naming conventions, preferred libraries, architecture decisions, or testing patterns. Every new chat session starts from zero — an amnesiac genius that writes excellent code in the wrong framework. Context engineering fixes that by building structured, version-controlled configuration files that teach the AI how your project works.

The customization system has seven distinct surfaces. Custom instructions activate automatically based on file paths. Prompt files create reusable slash commands for recurring tasks. Custom agents define specialized personas with restricted tool access. Skills package domain knowledge for cross-tool portability. MCP servers connect external tools. Hooks automate pre- and post-chat actions. Plugins extend agent capabilities from the marketplace. Each surface has a defined file format, trigger mechanism, and scope — and they compose together through a three-axis model.

In this course you build the complete picture. Starting with a blank .github folder, each lesson adds one layer of context capability — from always-on instructions through lifecycle hooks. By the end, you have a production-grade .github and /docs tree that makes Copilot behave like a project-aware expert. Every demo uses VS Code and the Copilot CLI.

1Understand why context engineering matters
Learn why AI output quality correlates directly with input context quality, how context engineering differs from prompt engineering, and why the shift to agentic AI demands a systematic approach.
2Master the .github directory structure
Explore the complete .github folder anatomy — six subfolders, each controlling a different aspect of AI behavior. Learn the instruction layering pattern, precedence rules, and VS Code settings that customize file discovery.
3Build all seven customization surfaces
Create custom instructions with glob-scoped activation, prompt files with dynamic variables, agent personas with tool restrictions, SKILL.md capability packages, MCP server connections, and lifecycle hooks.
4Apply the three-axis context model
Understand horizontal (agents and skills), vertical (instructions), and diagonal (prompts) context axes. Learn how they compose automatically and design layered architectures for multi-framework projects.
5Integrate context into your SDLC
Apply context engineering to planning, coding, review, testing, and CI or CD. Build role-based workflows that move from curated context to validated changes.
6Avoid common anti-patterns
Recognize and fix instruction bloat, monolithic rule files, duplicated linter guidance, over-privileged agents, and stale context that fights the AI instead of guiding it.
Who Should Join?

Prerequisites

What you need before starting

VS CodeGitHub CopilotGitTypeScript
Developers who use GitHub Copilot and want to move beyond one-off prompting to systematic, project-wide AI customization. Familiarity with VS Code and basic Git is recommended. No specific programming language expertise is required.
💻

Individual Developers

You use Copilot daily and want it to generate code that matches your project conventions without constant manual correction.

🏗️

Tech Leads and Architects

You want to encode architecture decisions, coding standards, and security policies into AI configuration that applies automatically for the whole team.

⚙️

DevOps and Platform Engineers

You manage CI or CD pipelines and developer tooling, and want to integrate AI context into existing automation and governance workflows.

🤖

Teams Adopting Agentic AI

Your team is moving from autocomplete to agents, and needs a structured approach to agent configuration, skills, and multi-agent orchestration.

Course Outline

9 lessons + 1 interview questions · ~95 mins

Each lesson builds on the previous one — follow them in order for the best experience.

1

Why Context Engineering Matters

Video Lesson·7 mins

Understand why AI-assisted engineering fails without durable project context and why context engineering is the most impactful skill for AI-assisted development.

2💻

Curate Project Context with .github and /docs/

💻Video + Code Examples·10 mins

Build the shared context layer that powers every AI interaction in your project by combining .github behavioral guidance with /docs knowledge context.

3💻

Design Instruction Architecture

💻Video + Code Examples·9 mins

Design a layered instruction architecture with repository-wide defaults, path-specific scoping, and the three-axis context model that keeps AI guidance precise and maintainable.

4💻

Planning Workflows with Prompts and Plan Agents

💻Video + Code Examples·10 mins

Separate planning from implementation using prompt files for repeatable workflows and read-only planning agents that decompose tasks, surface ambiguity, and produce actionable implementation plans.

5💻

Implementation Workflows with Agents and Skills

💻Video + Code Examples·11 mins

Design implementation workflows using custom agents for role separation, skills for reusable patterns, and TDD handoffs for disciplined execution — all governed by least-privilege tool boundaries.

6💻

MCP, Hooks, and Guardrails

💻Video + Code Examples·10 mins

Extend agent capabilities with MCP servers for external tool access, then enforce safety with hooks and validation scripts that run at key lifecycle points — keeping your AI workflows powerful and trustworthy.

7💻

Surface Strategy: Copilot, Claude, and Codex

💻Video + Code Examples·25 mins

Build a portable context layer that works across GitHub Copilot, Claude Code, and Codex by separating shared project knowledge from tool-specific wrappers.

8

Context Engineering Governance & Operating Model

Video Lesson·12 mins

Context engineering is the operating model for coding agents — not just better prompts. Learn the four-layer architecture, MCP as the universal adaptor, the F.I.T. rule for lean context, and the maintenance discipline that keeps your agent aligned as the codebase evolves.

9💻

End-to-End AI-Assisted SDLC Workflow

💻Video + Code Examples·12 mins

Synthesize the entire course into one end-to-end AI-assisted engineering workflow. Map shared context, planning, implementation, tooling, and maintenance into a curate → plan → build → validate delivery loop.

10📝

Course Quiz

📝Assessment·10 mins

Test your context engineering knowledge with scenario-based and concept-check questions covering the full curate → plan → build → validate workflow and the seven customization surfaces.

Instructor
Nilay Parikh

Nilay Parikh

Founder · LocalM · ErgoSum

Technologist with 20+ years of engineering experience and an ML/AI practitioner since 2010. Founder of ErgoSum (quantitative & equity research) and LocalM (AI-assisted SDLC). Currently focused on AI Platform Engineering, Agentic AI, and Context Engineering.