Custom Commands & AI Memory

Purpose: End the cold start. Give your AI memory that persists across sessions.

The Problem

Every AI session starts from zero. No memory of yesterday’s bug. No context about your architecture. No idea it already tried the wrong approach last week.

The Fix

Three things:

1. Commands File

Create .agent/commands.md — a single file where you define every command your AI responds to. Start with two:

  • start session — read memory files, check recent changes, load context
  • end session — save what was learned, flag what’s unfinished

Add more as your workflow evolves. push, review, check status — whatever you find yourself repeating. One file, all your commands, always discoverable.

2. Memory Files

Create .agent/memory/ — a directory where your AI writes down what it learns:

  • Anti-patterns — mistakes it made, so it never repeats them
  • Project context — architecture decisions, conventions, known issues
  • Journal — what happened last session, what’s next

Your AI reads these on every session start. The files grow. The AI gets sharper. Day 1 vs day 30 on the same repo is a completely different experience.

3. Connect Them

Your start session command reads the memory. Your end session command updates it. The commands drive the memory. The memory compounds.

How to Use This

Give this URL to your AI and say: “Set up custom commands and memory files for my project based on this.”

Your AI will figure out the right structure for your stack.

Created by Mats Ljunggren — from daily use across 17 repos.