> ## Documentation Index
> Fetch the complete documentation index at: https://docs.antigen.sh/docs/llms.txt
> Use this file to discover all available pages before exploring further.

# Quick Start

> Get from signup to your first vulnerability in five minutes.

Get from signup to your first vulnerability in five minutes.

<Steps>
  <Step title="Connect your infrastructure">
    Go to **Settings → Integrations** and connect your cloud provider. Start with AWS. Antigen creates a read-only IAM role that discovers your services, network topology, and security configuration.

    This builds your Asset Map, which the agent uses to reason about attack paths across your architecture rather than scanning endpoints blindly.

    You can skip this step and come back later. The agent will still work through external reconnaissance, but connecting your infrastructure first gives it a more complete picture of your attack surface.
  </Step>

  <Step title="Submit your targets">
    Go to **Settings → Targets** and add the domains you want tested.

    ```text theme={null}
    example.com
    *.example.com
    api.example.com
    ```

    To ensure responsible use, all targets require approval from your Antigen security expert before testing can begin. For your first run, we approve targets within minutes during business hours.
  </Step>

  <Step title="Run your first pen test">
    Once your targets are approved, your Antigen security expert will schedule your first engagement. Every run executes inside a hardware-isolated Firecracker microVM using cyber-capable models developed through our partnership with Anthropic. You will see results appear on the Vulnerabilities page as the agent works.

    If you use the SDK, you can also start a run programmatically:

    ```bash theme={null}
    npm install @antigen/sdk
    ```

    ```ts theme={null}
    import Antigen from "@antigen/sdk";

    const antigen = new Antigen(process.env.ANTIGEN_API_KEY);
    const targets = await antigen.getApprovedTargets();

    const agent = antigen.agent({
      model: "claude-sonnet-4-6",
      guardrails: "Stop on critical vulnerabilities.",
    });

    const run = await agent.run(targets);

    for await (const event of run) {
      if (event.type === "finding") {
        console.log(`[${event.severity}] ${event.title}`);
      }
    }
    ```
  </Step>

  <Step title="Review your vulnerabilities">
    Open the **Vulnerabilities** page. Each vulnerability includes:

    * Severity: critical, high, medium, or low
    * A description of the issue
    * Proof: the exact requests and responses that demonstrate it
    * A recommended fix

    If automated remediation is available, you will see a pull request linked to the vulnerability. Review it, merge it, and Antigen re-tests to confirm the fix worked.
  </Step>
</Steps>

## Next steps

* [Skills](/docs/sdk/skills) to teach the agent about your specific infrastructure
* [SDK overview](/docs/sdk) to run agents programmatically
* [Integrations](/docs/integrations/overview) to connect Slack, GitHub, Linear, and more