snailsploit[$]
⌘K live
services
kai aizen
independent
contact for scope

offensive security.
for the new stack.

Three disciplines. One researcher. The same methodology behind 23 CVEs, 5 Linux kernel patches, and the AATMF framework — applied to your systems.

get in touchread the frameworks
23 CVEspublished
5 patchesLinux kernel mainline
AATMF15 tactics · 240+ techniques
P.R.O.M.P.Tcompositional red-team grammar
Adversarial Mindspublished book
Hakin9 / PenTest Magcontributing author
01 · adversarial-ai

AI Red Teaming

Break it before someone else does.

End-to-end adversarial testing of LLM-based systems — from raw model APIs to production agentic pipelines. Built on the AATMF framework: 15 tactics, 240+ techniques, structured scoring with AATMF-R.

scope
→Prompt injection — direct, indirect, multi-turn, agentic
→Defense fingerprinting and bypass chain planning
→Tool-call poisoning and MCP/A2A trust boundary testing
→Memory injection and persistent implant validation
→Sandbox escape: Pickle RCE, DNS exfil, code interpreter abuse
→LLM safety judge blind spots and classifier evasion
→AATMF-R risk scoring per finding
deliverables
✓Technical writeup per finding
✓AATMF-R scored vuln ledger
✓Remediation guidance
✓Executive summary
Pricing on request.
All engagements include signed NDA and scoped rules of engagement.
02 · offensive-security

Advanced Penetration Testing

Systematic, methodology-driven, documented.

Full-scope offensive security assessments across Linux kernel, Kubernetes, container runtimes, cloud infrastructure, and web applications. The same methodology behind 23 published CVEs and 5 Linux kernel patches.

scope
→Network and infrastructure — cluster escape, RBAC abuse, privilege escalation
→Container runtime — breakout, symlink escape, capabilities abuse
→Web application — SQLi, SSRF, IDOR, auth bypass, CSRF chains
→Linux kernel — subsystem-level review and PoC development
→Cloud — credential exposure, misconfiguration chaining, lateral movement
→OSS library review — dependency chain audits, zero-day discovery
deliverables
✓CVE-ready technical report
✓PoC code where applicable
✓Attack-path diagram
✓Remediation roadmap
✓Executive summary
Pricing on request.
All engagements include signed NDA and scoped rules of engagement.
03 · human-layer

Social Engineering Assessment

The human layer is always in scope.

Targeted campaigns that test the intersection of human psychology and technical controls. Grounded in the P.R.O.M.P.T framework and the adversarial-psychology work behind Adversarial Minds.

scope
→Phishing — spear, whaling, pretexting via custom infrastructure
→Vishing — live social engineering with full documentation
→Physical + digital hybrid — badge access, tailgating, USB drop
→AI-assisted persona development and detection bypass
→Employee security awareness gap analysis
→Adversarial Minds methodology — cognitive bias exploitation
deliverables
✓Campaign report with success rates
✓Per-employee redacted results
✓Awareness gap analysis
✓Training recommendations
Pricing on request.
All engagements include signed NDA and scoped rules of engagement.
04 · engage

start a conversation.

Outline your environment and scope. I'll respond with a proposed approach, timeline, and fixed or T&M price. No boilerplate proposals.

contact →or find me on linkedin.com/in/kaiaizen
methodology behind every engagement
framework
AATMF v3.1 →
15 tactics · 240+ techniques · 2,150+ procedures
research
43 published pieces →
prompt injection, agent security, container escape, kernel
disclosures
23 CVEs + 5 kernel →
k8s, apache, oss, wordpress, linux mainline