Skip to content
0.security0.security0Docs
Research Preview

Your multi-model, self-improving
open-source cybersecurity team.

Install 0, connect your models, hack & fix your systems.

Choose the workflow for your target, then inspect its evidence and outcome—not just its exit code—before treating a lead as confirmed or a repair as verified.

Your taskStart here
Assess an authorized web or API targetScan workflows and scope files
Review source or a packagereview and audit
Reproduce a finding or check a repairVerification results and secure
Evaluate an AI endpoint or agent integrationeval and agent-assure
Investigate identity and access pathsEngagements
Hunt source variants, inspect binaries, or use kernel evidenceResearch workflows and 0verse prerequisites

secure brings repository investigation, reproduction, repair attempts, regression tests, and independent repair verification into one workflow. It runs on the current worker, not in a newly provisioned sandbox. Publication is opt-in, and a completed run can still contain blocked findings: inspect the individual outcomes before accepting a patch.

Use API Keys to establish a supported model connection, then choose models in the console or configuration. Role-model assignments can use different models supported by that provider or gateway. Child agents inherit the parent connection; they do not automatically discover or switch to another provider account. Model choice is not a claim of automatically optimal routing.

Build a Hackstore extension to add your own tools to 0, test them locally, and share them with other users.

The local CLI needs no 0cloud account when you use your own model connection. Hosted inference keeps tools in your environment. Managed security work has separate scope, access and commercial terms.