Integrate ProtoWall With Your Coding Agent

We shipped something small but useful today: an open-source integration guide that coding agents can read and execute directly.

If you use Claude Code, Cursor, GitHub Copilot, Windsurf, or any similar tool, you can now add ProtoWall origin secret verification to your project without writing the middleware yourself.

One prompt

Paste this into your coding agent:

Read https://raw.githubusercontent.com/protowall/integrate/main/AGENTS.md
and follow the instructions to integrate ProtoWall origin secret
verification into this project.

The agent detects your framework from your project's dependency files, picks the right middleware pattern, and places it where it belongs. It also sets up the environment variable placeholder.

That's it. No manual work.

What it actually does

ProtoWall already works without any code changes. The proxy handles authentication and NDA acceptance. Your prototype runs untouched.

The origin secret is an optional extra layer. It's a header that ProtoWall sends with every proxied request. You add a quick check in your app that rejects requests without it. This blocks anyone who tries to access your prototype directly, bypassing the proxy.

The integration guide covers Express, Fastify, Hono, Next.js, Nuxt, SvelteKit, Remix, Flask, FastAPI, Django, Rails, and Go. If you're using something else, there's a general pattern the agent can adapt.

Claude Code users

There's also a slash command you can install:

mkdir -p .claude/commands
curl -o .claude/commands/integrate-protowall.md \
  https://raw.githubusercontent.com/protowall/integrate/main/claude-code/integrate-protowall.md

Then just run /integrate-protowall in any project. It fetches the latest guide, detects your stack, and applies the changes.

Why we built this

Most of our users are building prototypes with AI coding agents. The whole point of ProtoWall is that it doesn't require code changes. But if you want the origin secret verification, you do need to add one middleware check. That felt like exactly the kind of task an agent should handle.

So we wrote the guide in a format agents understand: structured, framework-specific, with clear detection rules and copy-paste code. The agent reads it, does the work, and tells you what it changed.

Links

Full documentation is at protowall.app/docs/agent-integration.

The integration guide and Claude Code command are open source at github.com/protowall/integrate. PRs for additional frameworks are welcome.