Live reporting
As we find it
New assets, CVEs and techniques are tested as they appear. Critical paths are reported the moment they’re proven.
Why now
Under $3,000 in model tokens, one broken image parser opened paths into OpenAI, Slack, Meta, GitHub Enterprise, Shopify and more.
HOW IT WORKS
Hacktron proves the attack path, then helps you close it.
The living surface: the organisation-wide attack surface modelled, its connections questioned, attacked continuously, and one exploit path proven from entry to impact.
- Map: The Organisation-Wide Attack Surface. Hacktron maps the software environment an attacker can reach, creating a connected view of your organisation’s exposure.
- Test: Pathfinding. Hacktron traces the exposures, relationships and trust boundaries an attacker could use. Every connection becomes a question: is this the path to a breach?
- Test: Continuous Pressure. Every change to your environment and every newly disclosed technique triggers the relevant attacks again, keeping your validated exposure current over time.
- Prove: Impact & Fix. Hacktron validates one complete path from initial access to business impact.
WHAT YOU GET
Every path proven. Every fix verified.
We don’t stop at proving the path to compromise. We help you close it, and the whole class of attack behind it.
Validated kill chains
Every path
Every step of the chain, with the request, response and evidence that proves it.
Remediation and retests
Fix the cause
Once your team ships the change, we rerun the attack to confirm the path is closed.
Production-safe testing
Safe by design
Aligned, monitored agents work only inside the rules of engagement you agree.
WHY HACKTRON
Elite researchers direct. AI agents scale.
Our researchers find the complex, high-end bugs current models can’t, the kind only top human researchers catch.
- OpenAIInternal repos
- Google AntigravityRCE
- OpenAI AtlasBrowser takeover
- Perplexity CometUXSS
- SupabaseSupaPwn
- PostHogProd database
- Cloudflare vinextSession hijack
- VercelWAF bypass
- BeyondTrustPre-auth RCE
- Palo Alto GlobalProtectAuth bypass
- Ivanti EPMMPre-auth RCE
- Dassault Delmia AprisoPre-auth RCE
- OpenAMPre-auth RCE
- Metabase CloudMulti-tenant RCE
- Discord DesktopRCE
- Element DesktopRCE
- VS CodeRCE
- VS Code CopilotRCE
- CluelyTurned into malware
Zayne
Co-Founder, CEO
Cambridge CS. DEF CON CTF runner-up, credited with 15 CVEs. Security experience across government, military and TikTok.
Harsh
Co-Founder, Chief Research Officer
Former ProjectDiscovery researcher. Bug bounty hunter featured for hacking OpenAI, Apple, Atlassian, BeyondTrust, Meta, Slack and GitHub.
Mohan
Co-Founder, CTO
Former Cure53 senior researcher. Black Hat and DEF CON speaker. Built Electrovolt, a penetration testing firm, to €1.5M in revenue.
The agents
Hacktron AI agents
Every technique our researchers prove becomes a workflow. Aligned and monitored by those researchers, the agents run it against your whole environment, around the clock.
FAQ
Frequently asked questions.
What is continuous red teaming?
Red teaming that doesn’t stop at a test window. Researchers and AI agents attack your whole environment as it changes, chain what they find into real breach paths, and retest once your fixes land.
How is this different from a penetration test?
A pentest checks one scope at one point in time. Continuous red teaming runs across your whole attack surface and shows how weaknesses chain into a real breach.
How is it different from a scanner or attack-surface tool?
Those list assets and possible issues. We exploit them, connect the ones that chain together, and prove what an attacker could reach.
What’s in scope?
Your company-wide attack surface: external services, authentication, SaaS, dependencies and new features. We agree objectives and rules of engagement up front.
Will testing disrupt production?
No. Boundaries and prohibited techniques are agreed before we start, and our agents are aligned and monitored so they don’t take destructive actions. If anything risks availability, we stop and contact you on the agreed channel.
Does this replace our bug bounty program?
It complements it. We find the critical chains continuously, so fewer of them arrive as reactive bounty payouts.
Do you retest after we fix things?
Yes. Once your fix lands, we retest to confirm the same attack no longer works.
Start a Red Team engagement
Get hacked on your terms.
Talk to the researchers behind Hacktron about continuous red teaming across your whole attack surface.
Talk to our researchers · No commitment