Welcome aboard.
Two minutes to set up.

We'll create your account, give you the snippet to paste into Cursor or Claude Code, and watch your AI's first vulnerability get caught — together.

1 Create your account

One click. No password to remember.

We use your provider's account so we never see (or have to store) your password. Pick whichever you already use.

After sign-in we'll bring you back here to create your first project.

2 Name your first project

What are you building?

Just a name for your records. You can rename it anytime.

3 Connect your editor

Paste this once. Done forever.

Copy the snippet for the editor you use, paste it where it says, and your AI is now under our wing.

Cursor
Claude Code
Cline / Windsurf

Open Cursor → SettingsMCPAdd new MCP server. Paste this:

Then paste this into your project's .cursorrules or CLAUDE.md so the agent uses us properly:

Use the Literal Security tools for every code change:
- Call check_change before every Write/Edit
- Call check_dependency before every npm install
- Call generate_policies for every new database table
- Call pre_deploy_gate before any deploy
Never bypass a "block" decision without explicit user override.
Read the full prompt
4 Watch us catch a real vulnerability

Let's pretend your AI just wrote this.

We'll run the same check it would run if Cursor had asked us. Click the button below.

// api/notes.ts (what your AI tried to write)
const STRIPE = "sk_live_4eC39HqLyjWDarjt12345678901234";

app.get("/api/notes/:id", async (req, res) => {
  const r = await db.query(
    `SELECT * FROM notes WHERE id=${"${"}req.params.id}`
  );
  res.send(r);
});

🎉 You're set.

Your trial is active. Your editor is connected. Your AI now has a safety net.

Where to next?

Two things to do right now: