Open source tools for AI trust
Production-ready packages for detection, verification, and evaluation. AGPL-3.0 licensed.
12
Packages
2
Languages
Trust Intelligence Research
Building the science of reliable AI
Memory systems, agent reliability, adversarial robustness. Open research and open source for trust decisions in AI systems.
What We Study
Research Areas
Memory Systems
Field-theoretic memory architectures treating stored information as continuous fields governed by PDEs. Enabling persistent, coherent memory for AI agents.
Agent Reliability
Detection tools for strategic underperformance in AI systems. Activation probes, behavioral analysis, and sandbagging detection with 90-96% accuracy.
Adversarial Robustness
Evolutionary adversarial testing with quality-diversity optimization. Attack taxonomies, red-teaming frameworks, and multi-provider LLM targets.
Open Source
rotalabs-context
Context intelligence for AI agents - ingest, search, and subscribe to shared context.
rotalabs-steer
Runtime behavior control via activation steering vectors.
rotalabs-probe
Sandbagging detection via activation probes and behavioral analysis.
rotalabs-verity
Neuro-symbolic verified code synthesis with Z3 and CE2P feedback.
rotalabs-redqueen
Evolutionary adversarial testing with quality-diversity optimization.
rotalabs-eval
LLM evaluation with statistical rigor and confidence intervals.
rotalabs-cascade
Trust-based decision routing with 5-level verification cascade.
Publications
Recent Papers
2026
Field-Theoretic Memory for AI Agents
Memory architecture treating stored information as continuous fields governed by PDEs.
2026
Verity: Neuro-Symbolic Synthesis for Verified Code
CE2P translates formal verification failures into structured LLM feedback.
2026
Adversarial Testing for AI Systems
Evolutionary approaches to adversarial testing with competitive attack-defense pressure.
From the Blog
Latest Writing
March 01, 2026
Field-Theoretic Memory for AI Agents
We treat agent memory as continuous fields governed by partial differential equations instead of discrete database entries. The result: +116% F1 on multi-session reasoning and...
Read more
February 18, 2026
Shared Context for AI Agents: Introducing rotalabs-context
AI agents that can't share what they know make the same mistakes independently. We're releasing rotalabs-context - a context intelligence engine for ingesting, searching, and...
Read more
February 07, 2026
Statistical Rigor in LLM Evaluation
Why most LLM benchmarks are doing evaluation wrong, and how to fix it with confidence intervals, significance tests, and effect sizes.
Read more