Open to businesses · €50 starting credit
Security workspace: an AI security chatbot you supervise.
The LOKI security workspace is a chat-driven AI security workspace inside the LOKI client portal. You describe in plain language what you want tested. The assistant proposes concrete tool steps, and they run in isolated cloud sandboxes, not on your laptop. Before an active step runs you see the exact tool, target and command and decide whether it goes ahead. Every step lands in an audit log.
Only for websites you add yourself after confirming that you are authorized to test them. LOKI is notified of every site that is added.
You start with €50 credit. Every paid step shows its cost first.
- Scope
- Scan
- Findings
- Fix & retest
- Report
Test my site: shop.example.com
Scope noted. Running the first checks.
- TLS and certificate
- Security headers
- Exposed files
3 confirmed findings.
- HighBackup file readable at /backup.zipFixed
- MediumSession cookie without Secure flagFixed
- LowNo Content-Security-Policy headerFixed
Fixed all three. Test again.
Retest done. All three are closed.
How it works
From a sentence in chat to an approved, logged step.
01
Describe the test
Tell the assistant what you want checked, in your own words: for example the login of your customer portal or the headers of your public site.
02
Get a concrete proposal
The assistant turns your request into specific tool steps against the target in your scope. It proposes; it does not decide.
03
Review the exact step
For an active step the workspace shows the tool, the target and the literal command before anything runs. You approve or reject it. The step that runs is checked against the step you reviewed.
04
Run in a sandbox, keep the record
Approved steps run in an isolated cloud sandbox. The request, the decision and the result are written to the audit log of your organisation.
Supervised by design
What keeps a person in charge.
- Active steps wait for you
- Before an active step runs you see the tool, target and command and approve or reject it. Low-risk passive reads, such as HTTP, TLS and header checks, run straight away and are logged.
- Auto-approve is your switch
- You can let a conversation run every step its scope allows without asking. It is off until you turn it on, it does not widen the scope, and turning it on or off is itself written to the audit log.
- Isolated cloud sandboxes
- Tools run in an isolated cloud sandbox tied to your session, not on your own machines. A changed target or command is refused before it runs.
- An audit log of every step
- Proposed, approved, rejected and executed steps are recorded, with the decision and what came back.
- Encrypted per organisation
- Conversations, audit logs and the credentials of tools you connect are encrypted per organisation.
- Authorized scope only
- The workspace is for systems you own or are authorized in writing to test. You confirm that for every site you add, and LOKI is notified of each one.
How it differs
AI security chatbot, scanner or AI security workspace?
| General AI chatbot | Automated scanner | LOKI security workspace | |
|---|---|---|---|
| What it does | Answers questions and suggests commands | Runs a fixed set of known checks | Proposes concrete tool steps for your request and runs the approved ones |
| Where tools run | On your own machine, if you run the commands yourself | On the scanner's systems or yours | In an isolated cloud sandbox |
| Who decides what runs | You, by copying a command | A schedule or whoever starts the scan | You, per active step, with the exact tool, target and command in front of you |
| What is recorded | The chat history | A scan report | An audit log of every step and decision |
| Scope | Whatever is typed | The targets that are configured | Only the sites you added after confirming you are authorized |
The first two columns describe how these kinds of tools typically work; individual products differ.
What it is not
Clear limits.
- Not a human-reviewed pentest report. Tool output in the workspace is working material; findings checked by a person, with evidence, the fix and a retest, come from a SKYEN pentest.
- Not a certificate or a compliance statement.
- Not for consumers. Accounts are for businesses and organisations testing systems they are authorized to test.
- Not for systems you do not own or are not authorized to test.
Related
Where the workspace fits.
- SKYEN web app and API pentest
One human-reviewed pentest per agreed month, from €500 excluding VAT.
- Orvyn AI agent test
Agreed attack scenarios against your own AI agent or chatbot. Private preview.
- What does a pentest cost?
Price table by type of test, with sources.
Questions about the security workspace
What is a security workspace?
What is an AI security chatbot?
Can the assistant run anything without my approval?
Where do the tools run?
How is my data protected?
Which systems can I test with it?
How do I get access and what does it cost?
Does the security workspace replace a pentest?
Start with your own site.
Add a site you are authorized to test, tell the agent what you want checked and follow every step in the chat.
You start with €50 credit. Every paid step shows its cost first.