operator online58 classes11 agents0 auto-submitsscope-locked
The benchmark

Measured.Not claimed.

Every tool says it finds bugs. HUNTR gets graded — against targets where the answers are already known. “It works” is a number, not a vibe.

attack-class coverage58 / 58 classes
SQLiXSSSSRFIDORSSTIXXERCELFINoSQLiLDAPCORSCSRFDOM-XSSSAMLOAuthJWTGraphQLDeserializationPrototype-PollHost-HeaderHTTP-SmugglingCache-PoisoningOpen-RedirectClickjackingFile-UploadBusiness-LogicAuth-BypassMFA-BypassBrute-ForceCAPTCHA-BypassATOPassword-ResetSessionRace-ConditionAPI-MisconfigCloud-MisconfigShadow-APISPA-APISource-LeakSubdomain-TOTLS/NetworkWebSocketHTML-InjectionNTLM-InfoJS-AnalysisLLM/AIRAG-VectorError-HandlingASP.NETDjangoLaravelNext.jsNode.jsSpringSharePointK8sCI/CDgRPC
hunted end-to-end — recon to validated report
58
Vulnerability classes

Hunted end-to-end, recon to validated report.

11
Specialist agents

Recon, exploit, chaining, reporting — each a focused agent.

0
Auto-submissions

You review and send. Every finding, every time.

The method

No grading on a curve.

We don't estimate. HUNTR is run against known-vulnerable targets, so every hit can be checked against what's actually there — results, not a curated demo.

Local lab

Known-vuln lab

Black-box web + API lab. SQLi, XSS, SSTI, SSRF, IDOR, authz, CORS, LFI — answers known cold.

Auth lab

Access control

BFLA, broken access control, privilege boundaries. Tested grey-box.

ASP.NET

Framework depth

Proves the engine retools its methodology for the stack it finds.

Public

Real test sites

Industry-standard vulnweb targets, scored against known baselines.

Scoring

Signal vs. noise

Did it find the bug? Did it cry wolf? Money-weighted, FP-penalized.

Control

Deterministic

A no-LLM path makes core coverage reproducible — run to run.

Straight talk: HUNTR is pre-launch. These numbers are what the engine does on a controlled benchmark with certain ground truth — capability, not customer logos. Real disclosed findings land here as they're published. We'd rather show you a reproducible benchmark than a wall of logos we haven't earned.

Put it on a real target.