How it works
Authorized scope in, filed disclosure out
KinetixZero mirrors how careful research actually runs — four stages, with a human in the loop at the decisions that matter.
- 01
Define authorized scope
Everything starts with permission. You create a project, choose an authorization type, and attest to it.
- Set in-scope and out-of-scope boundaries for the engagement.
- The attestation is recorded with who attested and when.
- Scope travels with every finding and export for that project.
- 02
Analyze code & dependencies
Point KinetixZero at a Git repository or upload a source archive. It runs the analyzers read-only and records a scan.
- SAST with taint tracking across JavaScript, TypeScript and Python.
- Dependency resolution matched against OSV advisories.
- Redacted secret detection, with results surfaced by confidence.
- 03
Validate with the AI pass
Triage ranks the open findings. The AI layer assesses each one and shows cited evidence — you keep the verdict.
- Reachability is broken into controlled / reaches-sink / sanitized.
- Citations are verified against the exact code the model was shown.
- You confirm what's real and reproduce it; the AI never claims proof.
- 04
Disclose responsibly
Confirmed findings become a disclosure package, tracked to a deadline with a verifiable trail.
- Generate CVE 5.1, OSV, and PDF outputs from the finding's fields.
- Track vendor notification, acknowledgement, and fix to a deadline.
- A hash-chained audit log keeps the whole history auditable.