AI & me

Introducing AIME Directory - Your AI Development Companion

October 8, 2025

AI is transforming how we write code. If you’ve been using GitHub Copilot, Claude, or ChatGPT to help with development tasks, you know the potential. But you’ve probably also experienced the frustration: “How do I make this AI understand my project better?” or “Where can I find that perfect prompt I used last week?”

That’s exactly why I built AIME Directory.

The Problem with AI-Assisted Development

Let me paint a picture. You’re working on a React project. You want GitHub Copilot to follow your team’s coding standards. You need it to understand your project structure, your testing approach, your component patterns. But where do you put all this context? How do you organize it? How do you share it with your team?

Or maybe you discovered Model Context Protocol (MCP) servers that can extend Claude’s capabilities. But which ones are actually useful? How do you configure them? Where do you even find them all?

I’ve been collecting prompts, instructions, and configurations for months. Scattered across browser bookmarks, Notion pages, random GitHub gists. Every time I started a new project, I had to hunt down the same resources again. There had to be a better way.

Enter AIME Directory

AIME Directory is my answer to this chaos. It’s a curated platform that brings together everything you need for AI-assisted development in one place:

Model Context Protocol Servers (MCPs): Over 800 MCP servers that extend Claude’s capabilities. Want to connect Claude to your GitHub repo? There’s an MCP for that. Need to query your PostgreSQL database? Another MCP. Search through your codebase? You guessed it. Each entry includes the configuration you need to get started immediately.

Instructions: Framework-specific guidelines that teach AI assistants how to work with your stack. TypeScript best practices, React patterns, Node.js conventions, Angular architecture - all formatted as GitHub Copilot instruction files. Drop them in your .github/instructions/ folder and watch Copilot adapt to your project.

Prompts: Ready-to-use prompts for common development tasks. Code reviews, debugging sessions, documentation generation, refactoring tasks. No more starting from a blank prompt every time.

VSCode Configurations: Curated settings presets for GitHub Copilot and other AI tools. Want to enable agent mode? Need better chat settings? Looking for accessibility features? These configs are battle-tested and ready to copy into your .vscode/settings.json.

But Here’s the Real Magic

The directory itself is great. But what makes AIME Directory truly useful is the collection feature. Think of it as your personal toolkit that you can customize and export.

Here’s how it works: As you browse through the directory, you can add items to your collection with a single click. Found an MCP you want to try? Added. Discovered a useful instruction file? Added. Spotted a prompt that could help with your code reviews? Added.

Once you’ve built your collection, hit the “Export ZIP” button. AIME Directory generates a complete, ready-to-use project structure:

aime-collection-2025-10-08.zip
├── .vscode/
│   ├── mcp.json          # All your MCPs, configured
│   └── settings.json     # Merged VSCode settings
├── .github/
│   └── instructions/     # GitHub Copilot instructions
└── prompts/              # Your saved prompts

Drop this into your project, and you’re immediately set up with all the AI tooling you collected. No copy-pasting from 20 different tabs. No hunting for that configuration syntax. It just works.

Why I Built This

You might wonder: “Why not just bookmark these things?” Fair question. I tried that. The problem is context switching. When you’re deep in a project and need to configure an MCP server, you don’t want to:

  1. Search through your bookmarks
  2. Open the GitHub repo
  3. Find the configuration section in the README
  4. Figure out the syntax for your specific setup
  5. Copy it to the right file
  6. Hope you got it right

With AIME Directory, you:

  1. Search for what you need
  2. Add it to your collection
  3. Export
  4. Done

The directory handles the configuration format, the file structure, everything. You get back to coding faster.

Real-World Use Cases

Let me give you a few examples of how I use this:

Starting a New TypeScript Project: I collect the TypeScript instruction file, the ESLint prompt, the GitHub MCP (for repo context), and the Copilot Essentials VSCode config. Export, extract, commit. My AI tooling is set up before I write a single line of code.

Onboarding a Team Member: Instead of sending them a 5-page document about our Copilot configuration, I share a collection link. They browse, export, and have everything configured in minutes.

Experimenting with MCPs: I wanted to try different database MCPs for a project. Added SQLite, PostgreSQL, and a few others to a collection. Exported and tested each one. Found the winner, cleaned up the rest.

The Technical Side (For the Curious)

AIME Directory is built with Astro, which keeps it fast and lightweight. Content is managed through Astro’s content collections - every MCP, instruction, prompt, and config is a markdown file with frontmatter. This makes it easy to search, filter, and version control.

The export feature uses JSZip to build the archive client-side. No server needed, no privacy concerns. Your collection stays in your browser’s localStorage until you explicitly export it.

All content is indexed for search, tagged for filtering, and organized into logical sections. The site is static, which means it loads instantly and costs almost nothing to host.

What’s Next?

This is just the beginning. I’m continuously adding new MCPs as they’re released, creating more instruction files for popular frameworks, and curating prompts based on real development workflows.

If you’re building AI-powered software, AIME Directory is built for you. Whether you’re just starting with GitHub Copilot or you’re already deep into MCP servers and custom instructions, there’s something here to make your workflow better.

Check it out at aime.directory. Browse around, build a collection, export it, and see if it saves you time. I think it will.

And hey, if you have ideas for what should be in the directory, I’d love to hear from you. More on that in my next post about contributing to AIME Directory.

Happy coding! 🚀