Identity
Unverified · 0 matching sources · 2026-07-22No accepted supporting source is recorded for this field.
Agenova records RunSybil as a research agents product from runsybil.com. No capability, pricing, or deployment field currently has publication-qualified Evidence; those claims remain Unknown.
Agenova verified this product identity, but accepted official evidence is currently insufficient for specific capability, pricing, integration, deployment, privacy, or enterprise-control claims. Unverified fields remain Unknown.
RunSybil is an AI-native offensive security platform built around Sybil, an autonomous agent that continuously tests applications and infrastructure for exploitable vulnerabilities by reasoning about a system the way an elite human researcher would, across the entire stack, on every deployment. Rather than scanning for signatures, Sybil reasons like an attacker: it interacts with systems dynamically, explores authentication boundaries, finds and validates vulnerabilities, and chains them across layers to surface real exploitable paths, the way an application vulnerability becomes the entry point for full infrastructure compromise. The company frames its role as owning Phase 4 of CTEM, Validation: where most exposure-management programs stop at discovery and prioritization, Sybil continuously attacks applications and infrastructure with real adversarial reasoning to prove whether findings from other tools are actually exploitable, delivering pre-validated findings with zero triage burden. It positions itself as a replacement for both bug bounties (expensive, unpredictable, cherry-picked targets) and point-in-time pentests (stale the moment you ship), offering continuous coverage at predictable cost across application and infrastructure layers, testing for cross-tenant data access, privilege escalation, transaction manipulation, broken authorization, IAM misconfigurations, container escapes, CI/CD secret exposure, and lateral movement. RunSybil describes Sybil as the only platform running live exploitation with AI agents that learn about a target and get more efficient with experience; in one financial-platform engagement Sybil exploited a low-severity flaw, then used that knowledge to probe hidden endpoints and find a critical unauthenticated flaw exposing all customer data that black-box tools missed. Founded in 2023 by OpenAI's first security researcher and Meta's former offensive security lead with a team from Mandiant, NCC Group, and Trail of Bits, backed by a $40M Series A, RunSybil sits in the offensive cluster alongside XBOW, Horizon3, and Terra, differentiated by continuous per-deployment black-box testing that chains vulnerabilities across the full application-to-infrastructure stack.
No decision field has accepted public evidence yet.
0 of 8 decision fields supported. Evidence coverage is not a product score.
Known identity and product metadata appear below. The eight decision fields remain Unverified until accepted evidence supports them.
Request evidence →Documented product facts and Source-Backed claims do not establish execution quality or performance. Only publication-approved execution evidence can create an Execution-Verified state.
Documented product intelligence is not execution evidence or a performance claim. No publication-approved Execution-Verified edge appears unless its governed evidence passes.
No evidence-qualified capability edge is published for this profile. Provider-listed labels are not promoted as verified capabilities.
No accepted supporting source is recorded for this field.
No accepted supporting source is recorded for this field.
No accepted supporting source is recorded for this field.
No accepted supporting source is recorded for this field.
No accepted supporting source is recorded for this field.
No accepted supporting source is recorded for this field.
No accepted supporting source is recorded for this field.
No accepted supporting source is recorded for this field.
No accepted supporting source is recorded for this field.
These sources support only the fields named below. A missing source is reported as unknown, not as a negative product claim.