Input
Repo, scope, threat model
Your in-house team is already good. We give them coverage at the scale and pace AI permits, without replacing what they're already doing well. Findings come to you. Public bounty programs are for last-mile open call; we're the first-mile authorized review layer.
AI-driven source review across the contracts, services, and integrations you authorize. Methodology is documented at rainbowsix.dev; framework, bug-class library, and cite-or-KILL panel are auditable. You can read what we do before we do it.
Your engineers triage, decide, fix on your timeline. You own the disclosure. We're the coverage layer; you're the security team. We do not antagonize your existing audit firm or your in-house staff. We make their coverage budget go further.
We publish the framework openly where publication is safe. Every finding candidate ships with falsifier-first reasoning, local PoC tests where applicable, and explicit severity calibration. Your CISO can audit our process directly.
Current diligence packet: dated platform receipts, private responsible-disclosure records, and redacted local proof artifacts for account-abstraction, DeFi accounting, cross-chain replay, signer authority, and module-composition findings. Exact records are shared only inside authorized private review channels.
1 core contract or up to 5,000 LOC. Weekly review pass, monthly call.
Up to 25,000 LOC across multiple contracts. Daily review pass, weekly call.
Broad in-scope surface. Continuous authorized review plus quarterly tabletop exercises.
First 30 days as a design partner: free. After 30 days you decide: sign on at the tier above, or part as friends with the non-sensitive coverage report in your possession. Actual vulnerability details require an authorized private channel and mutual confidentiality.
Public and private work samples are available in redacted form under authorized diligence:
We do not publish live exploit detail, unresolved private disclosure detail, or customer-specific findings. The public page is the invitation; the evidence room is private.
Methodology source: github.com/CrunchyJohnHaven/rainbow-six-methodology (CC BY 4.0 + MIT).
If continuous AI red team coverage shaped like this fits your team, a 30-minute discovery call is the right next step. No theater, no pressure, just a working conversation about what coverage you need and how we'd fit alongside your existing team.
Contact: john@rainbowsix.dev
Founder: John Bradley, Rainbow Six Lab. Army Combat Engineers, Ranger-qualified, Sapper-tabbed. Built the methodology over multiple years; achieved breakthrough in 2026 by industrializing the AI feedback loop into a cite-or-KILL panel + audit-saturation calibration.