Introducing AREA 51: Open-source Email and HTTP Callback Infrastructure for Pentesters and AI Agents

Penetration testing often depends on infrastructure that goes beyond finding vulnerabilities. Testers need multiple email addresses, access to verification codes, somewhere to receive out-of-band callbacks, and a reliable way to share these resources across a team. And while none of this is especially complex, it can create a surprising amount of setup and coordination.

My colleagues and I at Thoropass built AREA 51 to eliminate this back-and-forth from our own pentests. It provides self-hosted email and HTTP callback infrastructure for pentesters and authorized AI agents. We’re now making it available to the security community as an open-source project.

Related: Why AI in Pentesting Must Stand up to Independent Scrutiny

‍

Email access without the overhead

Application penetration tests often require several user accounts to examine different roles, permissions, workflows, or account states. This has traditionally meant arranging aliases for multiple pentesters or asking the customer to create and approve each testing address. AREA 51 provides a catch-all inbox for a dedicated testing domain. Where customer-approved testing arrangements permit it, pentesters can create purpose-specific addresses without provisioning every alias individually. Messages are available through a central dashboard and can also be retrieved programmatically.

This supports account registration, password resets, magic-link authentication, email-based multifactor authentication, invitation workflows, and notification testing. 

‍

Supporting authorized AI agents

Email-dependent workflows can also interrupt AI-supported testing. An agent may be able to create a test account but still need a pentester to retrieve a verification code or open an activation link.

AREA 51 makes received emails available through REST and Model Context Protocol (MCP) interfaces. An authorized agent can retrieve codes, accept invitations, and complete other assigned workflows without relying on a pentester to relay information from an inbox. 

AREA 51 is infrastructure, not a replacement for professional oversight, so the pentester remains responsible for defining the task, controlling access, setting boundaries, and reviewing the results.

Related: Can AI replace pentesters?

‍

Capturing out-of-band callbacks

Some vulnerabilities don’t produce a visible response inside the application. Blind server-side request forgery and blind cross-site scripting, for example, may only be confirmed when the target sends a request to infrastructure controlled by the tester.

AREA 51 captures these HTTP requests and makes them available through the same MCP and REST interfaces used for email. Testers and agents can also configure custom HTTP responses and host test scripts, supporting scenarios in which the callback endpoint’s content or behavior needs to be controlled.

‍

Running within your Cloudflare account

AREA 51 runs within the user’s own Cloudflare account and uses a dedicated domain. Depending on usage, it can operate on Cloudflare’s free service tiers without separate servers or containers to manage.

After the initial configuration, the included setup command provisions the dashboard, storage, email routing, and callback infrastructure. This creates a reusable environment that remains under the testing team’s control.

As with any offensive security tool, AREA 51 should only be used within an explicitly authorized scope. Callback endpoints, hosted scripts, and agent access should all be covered by the engagement’s agreed testing arrangements.

‍

Built for practical testing workflows

AREA 51 began as a response to recurring operational problems during real engagements. It does not change the fundamentals of penetration testing, but it removes the small interruptions that prevent pentesters and their tools from working efficiently.

The project also reflects how we approach AI-enabled penetration testing at Thoropass. AI can execute defined workflows and reduce repetitive work, while experienced pentesters establish scope, apply judgment, interpret results, and remain accountable for the engagement.

Thoropass holds CREST Penetration Testing accreditation and was included in the first cohort of providers to earn CREST’s AI-Enabled Penetration Testing accreditation, which independently assesses how AI is governed and used within accredited services.

Explore the AREA 51 source code and testing playbooks, or follow the setup instructions to deploy it.

We hope it saves other offensive security teams some of the same setup and coordination headaches it has saved us.

‍