KinoSec.ai Becomes the #1 Black-Box Autonomous Pentesting Platform in the World

92.3% Success Rate — Surpassing XBOW and Most Human Pentesters

KinoSec has achieved the highest score in the XBOW Black-Box Benchmark, reaching a 92.3% success rate across 104 real-world security challenges.

This makes KinoSec the top-performing autonomous black-box penetration testing platform globally, outperforming XBOW itself (85%), leading open-source autonomous agents, combined human pentest teams, and individual pentesters at every experience level.

About the XBOW Benchmark

The XBOW benchmark is a rigorous black-box penetration testing evaluation consisting of 104 security challenges designed to simulate real-world attack surfaces. It measures:

• Exploitation success rate — can the system actually pop the vuln?
• Autonomous reasoning — does it adapt when the first approach fails?
• Vulnerability chaining — can it combine multiple weaknesses into a working exploit?
• Tool orchestration — does it select and sequence the right tools?

Critically, this is black-box testing: no source code, no internal documentation, no prior knowledge. Only the exposed attack surface — exactly what a real attacker sees.

Benchmark Results

KinoSec’s 92.3% score represents a 7.3-point lead over the next autonomous platform and outperforms even a combined team of human pentesters.

Results Summary

By Vulnerability Class

Per-Challenge Results

What This Means for Security Teams

Most organizations face a stark tradeoff: hire expensive senior pentesters who can only test periodically, or rely on scanners that miss anything beyond surface-level CVEs.

KinoSec eliminates that tradeoff. A platform that outperforms principal-level pentesters can run continuously, on every deployment, at a fraction of the cost of a manual engagement. This means:

• Continuous coverage — not quarterly snapshots that are stale before the report is delivered.
• Consistent depth — the same exploit-chaining and reasoning capability on every scan, without fatigue or time pressure.
• Faster remediation — findings surface in hours, not weeks, so your team can fix issues before attackers find them.

Methodology

• Testing Mode: Fully autonomous black-box penetration testing
• Source Code Access: None — agent interacts only via HTTP
• Max Turns: 50 per challenge
• Model: Claude Sonnet 4.6
• Infrastructure: Docker Compose challenges running locally

KinoSec ran the full 104-challenge XBOW benchmark suite under standard black-box conditions. No challenges were excluded, no parameters were tuned per-challenge, and no manual intervention occurred during execution. The same scanning pipeline available to every KinoSec customer was used for the evaluation.

All comparison figures are sourced from the XBOW benchmark leaderboard.

Try It Yourself

KinoSec is available today. Add your assets, run a scan, and see the same autonomous pentesting capability that topped the XBOW benchmark — applied to your own infrastructure.

Get started for free