MQL Benchmark
A 30,000-example open-source benchmark for evaluating natural-language → DSL generation, with a public model leaderboard.
Projects
Frameworks, agents, and open-source work.
Trust, Then Autonomy
A framework for evaluating earned autonomy in deployed AI systems.
Secure-by-Design Agents
How we architected ASA and ADÉ for adversarial production.
Evaluating LLM-Generated Detection Rules
A benchmark and three metrics for measuring LLM-generated cybersecurity rules — CAMLIS 2025.
BabbelPhish
Accelerating Adoption of Domain-Specific Languages with Large Language Models.
BabbelPhish Dataset
Open-source natural language to domain-specific language dataset for email security.
MalwareRL
Malware bypass research using reinforcement learning.