How to Use Configurable Chat Modes in VS Code — Copilot Just Got Way Smarter
TL;DR
GitHub Copilot in Visual Studio Code just introduced configurable chat modes — giving developers the power to create custom AI behaviors. Beyond the default Ask, Edit, and Agent modes, you can now define .chatmode.md files to make Copilot act exactly how you want. Whether you’re building SaaS products, reviewing code, or refactoring APIs, this update makes Copilot more adaptable, consistent, and team-ready.
🧠 What Are Copilot Chat Modes?
If you’ve used GitHub Copilot Chat, you already know it’s more than a code autocomplete tool — it’s an AI teammate living inside VS Code.
Chat modes determine how Copilot behaves.
Until recently, there were only three default modes:
- Ask Mode — great for quick Q&A or understanding code snippets.
- Edit Mode — lets you refactor or rewrite selected code.
- Agent Mode — gives Copilot broader workspace access for multi-file actions and automation.
Now, GitHub has made these modes configurable, meaning you can create your own chat modes that follow your workflow, coding standards, and tone.
⚙️ The Big Update: Configurable Chat Modes
Visual Studio Code now supports .chatmode.md files — simple Markdown configuration files that define:
- The description of the mode
- Which tools Copilot can access (like codebase, terminal, or tests)
- The AI model to use
- The specific instructions or persona Copilot should follow
This makes Copilot customizable like never before.
Example: You can create a mode called frontend-review.chatmode.md that instructs Copilot to act as a UI reviewer checking for accessibility, Tailwind standards, and component reusability.
Or a laravel-api-audit.chatmode.md mode to review controllers and suggest performance improvements.
🧩 Why This Update Matters
For startups, SaaS teams, and freelance developers, this feature is a serious productivity unlock:
✅ Consistent Code Reviews — Define rules once, and every team member gets the same AI-powered reviews.
✅ Faster Onboarding — New developers can use your pre-configured chat modes to learn standards quickly.
✅ Task-Specific Help — No more repeating “Hey Copilot, review this for security.” You just pick Security Review Mode.
✅ Scalable Workflows — Each project or microservice can have its own chat modes.
It’s like giving Copilot your team playbook — and letting it follow it perfectly.
🛠️ How to Create a Custom Chat Mode
Setting one up takes under 5 minutes.
- Open Command Palette → type “Chat: New Mode File”
- Save the file as
your-mode-name.chatmode.mdin your workspace - Add front matter (the metadata at the top):
---
description: "Review Laravel controllers for performance and code structure"
tools: ['codebase', 'search']
model: gpt-4-code
---
- Write your instructions below:
You are an expert Laravel reviewer.
- Check controllers for N+1 queries, redundant validation, and missing eager loading.
- Suggest improvements in bullet points.
- Keep answers concise and actionable.
- Open Copilot Chat → Choose your mode from the dropdown
- Start chatting!
Now Copilot acts exactly as you described — every single time.
💡 Real-World Examples
| Mode Name | Purpose | Example Use |
|---|---|---|
frontend-review.chatmode.md | Checks UI code for Tailwind best practices | “Review Dashboard.vue for design consistency” |
api-performance.chatmode.md | Audits Laravel controllers | “Find N+1 issues in UserController” |
security-scan.chatmode.md | Looks for unsafe input or SQL injection | “Check AuthController for security flaws” |
docs-writer.chatmode.md | Writes Markdown docs | “Document the new Invoice API endpoints” |
⚡ Pro Tips
- Keep modes simple at first — one focused purpose per mode.
- Store them in your repo so all teammates use the same versions.
- Use descriptive names —
nuxt-scaffold,laravel-audit, etc. - If your startup has specific naming or testing standards, encode them here.
- Pair this feature with GitHub Copilot’s Agent Mode for powerful automation.
🧭 Why Developers Love It
Configurable chat modes take Copilot from a “helpful assistant” to a custom team member that already understands your project.
For small teams or solo founders, it means less repetition.
For larger startups, it means predictable results and faster onboarding.
And if you’re a freelancer or agency (like TechGeeta 👀), it’s a huge differentiator — you can literally tell clients,
“We use pre-trained AI workflows for code review and development efficiency.”
That’s credibility + speed in one line.
🔮 The Future of AI Pair Programming
This update is a glimpse of what’s coming next — AI agents you can configure, share, and deploy across your dev team.
Soon, instead of asking “How do I do this?”, developers will say,
“Run Copilot in Startup Optimization Mode.”
And it’ll just do it.