02 / WEB Live view
RIGHT-SIDE LIVE SLOT / AUTO CYCLE 06S
Axel
Black-box testingDrives a real browser across your web app and returns evidence for every finding.
- Evidence
- Replayable browser evidence
- Operating mode
- Human checkpoints
Authorized AI security / Live evidenceAI penetration testing
AI penetration testing
with proof.
Not more noise.
Continuous security testing for websites, APIs, Android apps, codebases, and servers—run by six specialist agents while you control every boundary and inspect every finding.
Ownership verified Work visible live Evidence attached
OBSERVATORY / ACTIVE31 AUG · 21:47:06
V-27TARGET VERIFIED
NovaPassive
AxelLive view
MiraCross-device
KaiAPK ready
SableRepo aware
OrinRead only
Surfacecheckout.demo.elenxoai.com
Evidence184 linked objects
ControlHuman gate armed
LIVE FINDINGMEDIUM
Session token remains active after logoutReplay verified · browser evidence attached
Authorization validScope signed 08m ago
Specialist clock / 06 onlineNot one AI guessing.
Not one AI guessing.
A team finding.
Every six seconds the specialist clock advances. The agent arriving at the right-side live slot takes over the evidence brief.
Axel / Security-agent evaluationMeasured where security agents matter.
Measured where security agents matter.
CyberGym only.
One security benchmark keeps the comparison direct: the published CyberGym score for each listed runtime, presented on the same scale.
- 01
Axel
Purpose-built security agent
84.5/ 100PROMINENT USEComplete testing workflowWeb executionNative browserEvidenceConnected chainBoundaryBuilt-in gatesRemediationFinding-to-fix trace - 02
Claude Fable 5
General-purpose model · fallback enabled
83.8/ 100PROMINENT USEGeneral-purpose reasoningWeb executionTooling requiredEvidenceResponse contextBoundaryPrompt boundaryRemediationSuggested remediation - 03
GPT-5.6 Sol
General-purpose model
83.6/ 100PROMINENT USEGeneral-purpose agentic workWeb executionTooling requiredEvidenceResponse contextBoundaryPrompt boundaryRemediationSuggested remediation - 04
Claude Opus 4.8
General-purpose model
78.1/ 100PROMINENT USEGeneral-purpose analysisWeb executionTooling requiredEvidenceResponse contextBoundaryPrompt boundaryRemediationSuggested remediation
CyberGym scores only. Higher is better. Product-layer tooling, authorization policies, and deployment settings are outside this benchmark.
Autonomy with a visible boundaryThe agents move fast.
The agents move fast.
You hold the line.
Prove ownership before a scan starts. Watch the work. Approve anything that could cross the boundary.
CONTROL PLANE / RUN 7F2APOLICY ENFORCED
Ownership proved
Signed target scope accepted
Agents observed
Every action streamed live
Human checkpoint
Potential impact requires you
Audit sealed
Decision and evidence linked
policy://active-scanPause before impactimmutable log 184/184
Ownership first
No active testing begins until the target is verified.
Human checkpoints
Potentially disruptive actions pause for your explicit approval.
Rate limited
Every run stays inside controlled request and concurrency ceilings.
Traceable by design
Actions, decisions, evidence, and findings stay connected.
Signed authorization Controlled rate limits Safe expiry Traceable decisions
Live catalog / No surprise tiersStart with Nova.
Start with Nova.
Scale when evidence asks.
Four passive reconnaissance checks are included every month. Deeper specialists unlock when your scope needs them.
CATALOG STATE DATABASE SYNCEDUSD · monthly · credits reset each cycle
01FREE
Free
Essential reconnaissance for getting started.
$0/month
- 4 credits each month
- 1 project
- Nova passive reconnaissance
- Community support
4credits1projects1concurrent
Start with Nova 02STARTER
Starter
A practical toolkit for shipping teams.
$19/month
- 12 credits each month
- All specialist agents
- 2 targets per project
- Email support
12credits1projects1concurrent
Choose Starter 03MOST CHOSEN
Pro
Continuous security coverage for growing products.
$59/month
- 40 credits each month
- 3 active projects
- All specialist agents
- Priority support
40credits3projects1concurrent
Choose Pro 04BUSINESS
Business
Higher capacity and control for security teams.
$149/month
- 100 credits each month
- 10 active projects
- 2 concurrent scans
- Priority support
100credits10projects2concurrent
Choose Business Failed agent runs are refunded automatically. Free is web-only and Nova-only; paid tiers unlock all specialist agents and scan depths.
Supported technology / Connected evidenceSystems your team
Systems your team
already ships.
Bring repositories, Android packages, web journeys, server context, and compatible evidence storage into one accountable workflow.
PROOF BEFORE PRAISETrust should be inspectable.
Every result linked
Requests, responses, screenshots, code paths, and decisions remain attached.
Authorization first
Ownership and scope are established before active testing begins.
Systems stay yours
Integrations support the workflow without being presented as endorsements.
Early-access voice / Presentation previewWhat early security teams value.
Share verified feedback What early security teams value.
Evidence they can act on.
These cards are illustrative early-access feedback used to demonstrate the review experience. Replace them with approved customer quotations before public attribution.
“The connected evidence trail turns a security conversation into work an engineering team can actually pick up.”
“Visible boundaries and approval checkpoints make autonomous testing much easier to introduce responsibly.”
“Requests, browser actions, findings, and remediation context finally read as one continuous investigation.”
“Specialist agents map more naturally to the work than another general-purpose security chat window.”
Illustrative presentation copy—not published customer endorsements. Identity and permission are required before attribution.
Human channel / Scope firstSet the boundary.
PHONE+1 716 6630339Product and scope conversations EMAILinfo@elenxoai.comProduct questions and private security coordination
Set the boundary.
Then start the work.
Bring the target type, authorization context, and outcome you need. We will establish safe scope before active testing.