STACK.md — The Context File That Makes Claude Code Work

Overview

What is STACK.md?

STACK.md is a plain-English markdown file that SureStack generates and maintains for your project. It contains everything an AI needs to pick up exactly where you left off — what you've built, what's broken, what to do next, your architecture decisions, file paths, and technical details.

Project state — what's built, what's in progress, what's broken
Architecture — tech stack, services, file structure, deployment details
Decisions — why you chose certain approaches over others
Next steps — what to tackle next, ordered by priority
Known issues — bugs, limitations, and things to watch for

Workflow

How you use it

There's no CLI, no plugin, no integration. You copy your STACK.md from the SureStack app and paste it into any AI tool. That's it.

Copy from Phase 1

Open your project in app.surestack.dev. Hit "Copy project memory." Your STACK.md is now on your clipboard.

Paste into your AI

Open Claude, ChatGPT, Gemini, Cursor — whatever you use. Paste the STACK.md as your first message. Say "Here is my project context" and start building.

Build with full context

Your AI immediately knows your project. No setup questions, no explaining your stack. Your first prompt can be a feature request, not a configuration briefing.

Update when you're done

At the end of your session, SureStack gives you a closing prompt. Your AI outputs an updated STACK.md. Paste it back into SureStack. Next session starts where this one ended.

No lock-in

Why it works with every AI tool

STACK.md is plain markdown. Any AI that can read English can understand it. There's no special format, no proprietary schema, no SDK to install. If you can paste text, you can use SureStack.

Claude & Claude Code

Paste it as the first message, or save it as a CLAUDE.md file in your project root for automatic loading.

ChatGPT & Gemini

Paste it at the start of any conversation. Works in the browser, API, or any wrapper.

Cursor, Windsurf & IDE tools

Save the STACK.md as a file in your repo. These tools read project files automatically.

The loop

The session loop

Every build session follows the same three-phase loop. This is how your project memory stays up to date without you writing anything manually.

Phase 1 — Start

Copy your STACK.md from SureStack. Paste it into your AI tool. Start building.

Phase 2 — End

Before closing your AI session, copy the closing prompt from SureStack. Your AI outputs a complete, updated STACK.md reflecting everything you built.

Phase 3 — Save

Paste the updated STACK.md back into SureStack. Your project memory is now current. Next session starts with full context.

The result: your AI always starts informed. Your project never loses context between sessions. You never have to re-explain your stack.

The difference

Without vs. with STACK.md

The same prompt, two completely different starting points.

Without STACK.md

Add a /api/billing/portal endpoint that creates a Stripe customer portal session

I have a few questions before I start:

1. What framework are you using?
2. Where are your Stripe keys stored?
3. Do you have an existing Stripe integration?
4. What's your auth setup?
5. Which database stores user data?
6. What's your deploy URL?

With STACK.md

Add a /api/billing/portal endpoint that creates a Stripe customer portal session

I'll create the endpoint using STRIPE_SECRET_KEY from your env, authenticate via Clerk middleware, look up the Stripe customer ID in Supabase, and return the portal URL. Let me write that now.