AI security testing.
Evidence you
can follow.

Move from an authorized scope to recorded evidence, findings for human review, and a report. Bring your own model and executor. Keep the work in one place.

Free to start. Your model usage and infrastructure costs are yours.

Security workspaceIllustrative example
Task · DEMO-01Review the demo applicationScope and executor selected
Evidence recordedDEMO-P42

One observation. Its original execution.

Readable report. Traceable proof.Original evidence stays linked through retests.

Your model & infrastructure

Execution you can inspect

Human review before confirmation

Follow the work.
From scope to retest.

Chat starts the conversation. Tasks organize the work. Evidence gives you something to review.

Demo application reviewInteractive example · no live execution
SCOPE / DEMO-01Authorized target

A defined starting point.

Describe what you are authorized to review. Configure your model now and select a connected executor when you are ready to run.

Target
https://api.demo.test
Task
Review the demo application
Model
Your configured model profile
Execution
Your selected Linux or Windows executor

This is fictional sample data. No target is contacted.

Step 1 of 6

Your machine.
A temporary connection.

The agent connects your Linux VPS or Windows machine to Vulnetic. Commands run on the executor you select, using its installed tools and network access.

Set up your workspace
Your executorLinux / Windows
Outbound connectionExpires or can be revoked
VulneticYour workspace
Choose an access window
3 hours, 24 hours, 3 days or 7 days. Reconnection never extends the original deadline.
Run without elevated permissions
Linux uses a dedicated user without sudo; administrator access is needed for installation. Windows runs in an open terminal without administrator rights.
Reconnect within the authorization
Linux reconnects after a restart while access is valid. Windows can reconnect while the terminal stays open; closing it or restarting Windows requires a new authorization.
End access when you need to
Revoke the executor from the workspace. Expired access needs a new authorization before execution can continue.

Bring the model.
Connect the executor.
Make the work visible.

For independent security professionals and small teams who want a clear workflow around their authorized assessments.

Create your free account

Open your workspace and save your own model profile. You can configure it before connecting an executor.

Connect a Linux or Windows executor

Choose an access window and run the generated installation or terminal command on a machine you control.

Start with a scope you are authorized to test

Select the task's model and executor. Follow progress, inspect recorded tool results, review findings and request a retest.

Know where your data goes.

Remote execution does not mean everything stays on your machine.

On your executor

Commands, installed tools and new uploaded attachments in the executor's temporary workspace.

In Vulnetic

Account settings, task history, execution evidence and reports. Original attribution is retained.

With your model provider

Model requests can include task content and tool results. Review your provider's data handling before running an assessment.

Before you start.

A few practical answers.

Is Vulnetic free?

The hosted platform is free to start. You supply your own model access and execution infrastructure, and pay those providers directly for any usage. Free platform access does not include model tokens or a VPS.

Do I need a VPS to configure my model?

No. You can save and activate your model profile first. A connected, authorized executor is required when you test that connection or start work that needs remote execution.

What can the agent access?

Commands have the permissions and network reach of the agent's operating system user. Linux uses a dedicated account; the Windows terminal uses your current non-administrator account. Use a dedicated environment containing only the resources needed for your authorized scope.

Does every AI-generated finding count as confirmed?

No. Candidates, evidence and confirmation are separate. Model-generated text alone is not proof. Findings require evidence review and human confirmation; a report preserves their actual review status.

What happens when executor access expires?

Execution stops being available under that authorization. Reconnects and platform restarts do not renew its deadline. Task history, recorded evidence and reports remain in Vulnetic. Connect a newly authorized executor before continuing.

Can I use Vulnetic on any target?

Use it only for systems you own or have explicit permission to assess, within the agreed scope. Platform protections and account permissions apply to every execution.

Start your next review
with a clearer record.

Your model. Your executor. Evidence you can follow.

Create free account

Bring your own model and executor.