Stop Explaining Your Codebase to Your AI Every Time

Every conversation with your AI starts the same way. "I'm building a Rails app, deployed on Hetzner, using SQLite..." You've typed this a hundred times.

Your AI is smart. But it has no memory. Every chat starts from zero. Your project context, your conventions, your past decisions — gone.

What if your AI already knew all of that?

Here are five notes that make that happen.

1. Your stack, saved once

Write one note with your tech stack, deployment setup, and conventions. Now every conversation starts with context.

Now ask: "Write a background job that syncs user data to Stripe."

Your AI reads the note. It knows it's Rails, knows you use Solid Queue, knows your conventions. No preamble needed.

2. Error fixes you'll hit again

You spend 45 minutes debugging a Kamal deploy. You find the fix. A week later, same error. You Google it again.

Save the fix instead. Next time it happens: "I'm getting a health check timeout on Kamal deploy. Have I seen this before?"

Your AI finds the fix immediately. Thirty seconds instead of forty-five minutes.

3. Architecture decisions with reasoning

You chose SQLite over Postgres. You went with Turbo instead of React. There were reasons. Three months later, you can't remember them.

Title: Why SQLite over Postgres

Single-server deployment on Hetzner. No connection pooling needed.
Litestream for backups to S3.
Simpler ops. One fewer service to manage.
Will revisit if we need read replicas or concurrent writes beyond
what WAL mode handles.

When someone asks "why not Postgres?": "Summarize our database decision and the reasoning."

Done. No digging through old Slack threads.

4. Project rules your AI follows

Every codebase has unwritten rules. Where things go. What patterns you prefer. What to avoid. Your AI doesn't know any of them — unless you write them down.

Title: Project Conventions

Service objects live in app/services/.
Use Minitest, not RSpec.
Prefer Turbo Frames over custom JS.
No background jobs unless clearly needed.
All API responses use JSON serializers.
Keep controllers thin — logic goes in models or services.

Now when you ask your AI to write code, it follows your patterns. Not generic Stack Overflow patterns.

5. Code you'll reuse anyway

Every developer has code they reuse. Config patterns. Deployment scripts. Regex that took an hour to write.

Title: Snippet — Rack::Attack rate limiting

# config/initializers/rack_attack.rb
Rack::Attack.throttle("login/ip", limit: 10, period: 60) do |req|
  req.ip if req.path == "/session" && req.post?
end

Next project: "Show me how I set up rate limiting last time."

Your AI finds the snippet. You paste and adapt. No digging through old repos.

How this works

Your AI can read your notes through MCP. It's a protocol that connects AI to external data. Hjarni is a note app with a built-in MCP server.

No embeddings setup. No extra system to manage. Just notes your AI can read.

Connect it in two minutes:

1. Sign up at hjarni.com
2. In Claude or ChatGPT: add https://hjarni.com/mcp as a connector
3. Start asking questions only your notes can answer

Free tier gives you 25 notes with full MCP access. Ten notes in, your AI stops being a stranger.

New to Hjarni? Start with How to Give Claude Long-Term Memory.