Skip to main content
Verity180Governed autonomous security research
ProductSecurityCompareControlled pilot
Sign inRequest pilot
ProductSecurityCompareControlled pilotSign in
Request controlled pilot

platform architecture

An autonomous AI red team with the controls your security team needs.

Verity combines multi-agent security research, bounded exploit proof, white-box source audit, independent evidence verification, human approval, and remediation retesting in one authorization-bound platform. Verity Hunter is the customer-owned testing module; Verity Inbound Reports handles reports from outside researchers. Public corpora and controlled fixtures measure detector quality during development; they are not evidence of a live bounty, provider acceptance, or autonomous-bounty certification.

Request a controlled pilotOpen Hunter console
dual-lane execution spine3-gate enforced
Owned-Asset DAST Lane13 execution stages

Six behavioral HTTP finders, with authorization-bypass and SSRF handled as separately gated source-assisted classes.

BOLA (IDOR)Auth BypassSSRFSQLi and SSTIJWT Confusion
White-Box Source Audit Lane7 gVisor stages

Opt-in AST and cross-module analysis; canary proof and patch synthesis require separately verified Linux KVM and gVisor isolation.

VERITY_CANARY_*Safe-Sibling DiffKiro Prism CriticAuto-Retest
Principal isolation: owned_asset vs bounty_programSHA-256 Manifest

Multi-agent research, not signature scanning

Agents examine source, application behavior, API contracts, browser flows, prior evidence, and program rules; form hypotheses; compare models; and use only reviewed skills. Every hypothesis must still survive executed proof.

Verified assets and explicit scope

Bind each run to current ownership verification, normalized scope, and an expiring authorization grant pinned by SHA-256 digest.

Live-target and source-audit lanes

Test running web applications and APIs, or audit connected Python and TypeScript repositories you own — with the same execution-gated proof standard on both.

Authenticated two-actor web and API testing

Map customer-provided actor and victim identities to object-level and role-level authorization paths without storing raw credentials.

Bounded non-destructive proof and isolated source audits

Execute approved read-only plans through rate, request, time, cost, transport, and kill-switch gates. The opt-in source_audit lane remains disabled unless a compatible Linux KVM host proves the required gVisor boundary.

Independent evidence review

The production contract promotes only target-reaching proof that passes canary extraction, safe-sibling negative controls, cross-family critic review via Kiro Prism, and duplicate checks. Deployment and authorization evidence still determine whether a cell may run.

Patches verified by the exploit

Synthesize a minimal unified diff fix and replay the original proof-of-concept against it. Verified means the exploit stops reproducing and the project's tests still pass.

Remediation ownership and retest

Assign the issue, track engineering progress, and replay the original proof under current authorization to confirm closure.

Inbound Reports intake

Reports from external researchers, customers, partners, and internal teams enter the same governed evidence, remediation, and retest workflow as Verity's own findings.

Audit-ready reports

Generate immutable PDF and JSON reports tied to authorization grants, runs, findings, SHA-256 evidence manifests, and retest outcomes.

Verity180Governed autonomous security research

The operating system for autonomous security research: authorization-bound testing, independently verified findings, defensible evidence, and retest.

Product
  • Product
  • Security
  • Compare
  • Controlled pilot
Unified console
  • Hunter overview
  • Inbound Reports
  • Sign in
© 2026 Verity Inc. All rights reserved.Authorized scope, verified evidence, human interlocks