01
Map
Understand the architecture, state transitions, privileges, and trust boundaries.
AUTONOMOUS SECURITY RESEARCH
Security intelligence for the onchain world.
INVARYN is an agentic security researcher designed to investigate the full blockchain stack, form attacker driven hypotheses, and validate critical vulnerabilities through reproducible exploit paths.
RESEARCH SURFACE04 DOMAINS
- 01Smart contracts
- 02Blockchain infrastructure
- 03Zero knowledge systems
- 04Cryptographic protocols
WHY INVARYN / 001
THE SECURITY GAP
The hardest vulnerabilities do not respect boundaries.
Protocol security is not confined to a single contract or language. Failures emerge between components, assumptions, and trust boundaries. INVARYN is being built to investigate those interactions as one connected attack surface.
THE RESEARCH LOOP
From architecture to reproducible proof.
A structured investigation process designed to remove weak hypotheses before they become noisy findings.
02
Hypothesize
Form attacker driven paths across functions, contracts, and system components.
03
Reproduce
Turn promising paths into executable tests and controlled exploit conditions.
04
Validate
Confirm reachability and impact, then preserve the evidence needed to reproduce it.
RESEARCH SURFACE
One system. The full onchain stack.
Domain depth matters. Cross domain reasoning matters more when the failure lives between them.
Smart contracts
State transitions, economic logic, permissions, integrations, and adversarial composition.
Blockchain infrastructure
Consensus, networking, clients, bridges, cross chain state, and operational trust.
Zero knowledge systems
Constraint integrity, circuit assumptions, prover and verifier boundaries, and implementation risk.
Cryptographic protocols
Protocol design, key handling, authentication, signatures, and adversarial edge cases.
DESIGNED OUTPUT
Evidence, not a list of model guesses.
The goal is not to generate more alerts. It is to deliver a finding that a security team can understand, reproduce, and act on.
- Root cause and affected state
- Attacker path and required conditions
- Impact with explicit boundaries
- Reproducible proof of concept
INV / FINDING MODELVALIDATION PATH
C
CRITICAL / CROSS COMPONENT
Verified exploit path
- ROOT CAUSE
- Trust boundary violation
- REACHABILITY
- Attacker controlled
- PROOF
- Reproducible test
Conceptual output model. Product interface is in development.
RESPONSIBLE DISCLOSURE
Research will be published when it is safe to publish.
Technical reports and vulnerability research are currently moving through private research and coordinated disclosure processes. Public material will appear here as remediation and publication windows close.
Visit research ↗LAUNCHING SOON