Next Kick Labs

AI agent authorization: enforcing delegated access limits

Delegated AI access needs checks at execution, shared limits across concurrent work and clear withdrawal rules. Token exchange alone does not enforce…

Nov 4 • Fernando Lucktemberg

1

October 2026

AI agent identity: establishing trust across organizations

AI agent identity depends on trusted issuers, verified operator relationships and current credentials. The design covers discovery, evidence and…

Oct 21 • Fernando Lucktemberg

1

AI agent delegation security: Enforcing access limits

AI agent delegation needs enforced access limits, task ownership and recovery rules. Protocol support alone does not prove a deployment enforces them.

Oct 7 • Fernando Lucktemberg

September 2026

Running the threat model on FinBot

Running STRIDE, MAESTRO, OWASP and MITRE ATLAS against one agentic application, and the twelve questions its architecture raises but cannot answer.

Sep 23 • Fernando Lucktemberg

1

Threat modeling OWASP FinBot as an agentic application

Disclaimer

Sep 9 • Fernando Lucktemberg

1

July 2026

OWASP FinBot: An Agentic AI Security Test Range

OWASP FinBot is an intentionally vulnerable multi-agent app that scores exploits on real business state, turning agentic AI security claims into…

Jul 22 • Fernando Lucktemberg

Things I got wrong about agent security

Agent security frameworks name risks in full but rarely validate the defense. Threat modeling and infrastructure matter more than this month's…

Jul 15 • Fernando Lucktemberg

1

2

5

Q2 2026: 184 Subscribers, a Solo Quarter, and the Number That Actually Matters

An honest accounting of Q2 2026: 184 email subscribers, 728 followers, a solo quarter with no viral moments, and the Agentic AI Security Stack…

Jul 8 • Fernando Lucktemberg

1

June 2026

AI Security Testing as a Chain of Trust

AI security testing becomes credible when benchmarks, scanners, and guardrails compose into a chain of trust validated by application runtime telemetry.

Jun 24 • Fernando Lucktemberg

4

There is no NMAP for LLMs yet

LLM security tools produce different evidence. This lab shows how garak, promptfoo, DeepTeam, and Augustus should be scoped and interpreted.

Jun 17 • Fernando Lucktemberg

1

CyberSecEval on a Consumer GPU: What My Local Setup Could Actually Measure

A local CyberSecEval lab shows why response fields, harness state, and server configuration matter as much as model scores.

Jun 10 • Fernando Lucktemberg

1

In AI Vulnerability Research, the Pipeline Is Becoming the Product

Open-source AI vulnerability research tooling now covers discovery, proof construction, patching, triage, and evaluation, but verifiable pipelines…

Jun 3 • Fernando Lucktemberg

1