EngagementsThree ways to
Three ways to
work together.
Start free with the open-source harness. Bring me in for a fixed-scope audit when your agent is about to touch money, customer data, or production systems. Keep me on retainer when it never stops changing.
agentic-redteam
Open-source red-team harness you can run against your own agent — jailbreaks, injection, PII exfiltration. Publishing shortly.
FreeApache 2.0
Self-serve — clone and run
- Target-agnostic HTTP runner
- Jailbreak signature suite
- Output harmful-compliance detection
- PII & code-safety assertions
Most teams start here
AI Agent Security Audit
A fixed 1–2 week engagement that tells you exactly where your agent breaks and what to do about it.
$7,500 – $12,500fixed scope
1–2 weeks · delivered by me, not a junior
- Threat model mapped to OWASP LLM Top 10
- Red-team run against your endpoint
- Guardrail gap analysis + sample config PR
- Severity-ranked findings & remediation blueprint
- 60-minute executive debrief
Guardrail & Red-Team Retainer
Continuous assurance for teams shipping agent changes every week. Usually starts after an audit.
$4,500/ month
Rolling — 30 days' notice, no lock-in
- Red-team sweeps on every release
- Guardrail & eval maintenance as threats evolve
- Regression telemetry so safety can't drift
- Direct access for architecture review
Prices are USD for global engagements. Regional rates apply for India and the GCC. One or two design-partner discounts are available in exchange for a public case study — ask.