I Built a VS Code Extension to Turn Prompt Chaos into a Structured Workflow

Like many developers working with AI daily, I hit the same wall over and over:

• Prompts scattered across notes, docs, and chats
• Rewriting the same instructions repeatedly
• Losing “that one perfect prompt” I used last week
• No real system for chaining ideas into something more powerful

So I built Lumra — and more specifically, a VS Code extension that brings structured prompt management directly into the development workflow.

Explore it here: Lumra
VS Code Extension: Lumra VSCode Extension

The Core Idea

Instead of treating prompts as disposable inputs, Lumra treats them as reusable, structured assets.

Inside VS Code, you can:

• Store and organize your prompts
• Instantly reuse them without copy-paste chaos
• Build chains of prompts for more complex workflows
• Access everything without leaving your dev environment

Why This Changes the Workflow

1. Your Prompts Become a System

You stop thinking:

“What was that prompt again?”

And start thinking:

“Which prompt module should I use here?”

This shift alone removes a lot of friction.

2. Faster Iteration with Copilot

When combined with GitHub Copilot, things get interesting:

• Use Lumra to inject structured prompts
• Let Copilot generate outputs faster and more consistently
• Iterate quickly without rewriting context every time

You’re essentially giving Copilot better instructions, instantly.

3. Prompt Chain Planner = Better Outputs

One of the most powerful features is the Prompt Chain Planner.

Instead of writing one big messy prompt, you:

• Break problems into steps
• Design structured chains
• Reuse and refine each part

This leads to:

• More predictable outputs
• Higher quality responses
• Easier debugging of prompt logic

4. Agent-Friendly by Design

If you're experimenting with AI agents or multi-step workflows:

• Prompts are modular
• Chains are reusable
• Logic becomes composable

This makes Lumra a solid foundation for building agent-like systems without overengineering.

5. Everything Inside VS Code

No more context switching.

• No external docs
• No lost prompts
• No friction

Just:

Write code → Trigger prompt → Get result → Iterate

What It Feels Like in Practice

A simple flow becomes:

1. Open VS Code
2. Select a saved prompt or chain
3. Run it instantly
4. Combine with Copilot
5. Refine and save improvements

Over time, you build a personal prompt library that actually compounds in value.

Why I Built This

I didn’t want another prompt library.

I wanted:

• A developer-first workflow
• A way to scale prompt usage
• A system that makes AI feel like a real tool, not a hack

Would Love Feedback

If you're building with AI, experimenting with agents, or just tired of messy prompt workflows, I’d love to hear what you think.

Try it here: Lumra
Install extension: Lumra VSCode Extension

What’s your current prompt workflow like?