Technical
documentation

Core guides for understanding and using the framework

These guides provide the foundational knowledge for working with the Failure-First Embodied AI framework. They cover methodology, data structures, taxonomy design, and evaluation approaches.

Methodology

Failure Taxonomy Guide

Authoritative guide to the dual-taxonomy model and failure-first philosophy.

Attack Technique Evolution

Historical timeline of jailbreak techniques from 2022 to present.

Data & Datasets

Dataset User Guide

Practical instructions for researchers using the datasets.

Dataset Selection Guide

Decision tree for choosing the right dataset for your research.

Taxonomy

Comprehensive Scenario Classes

Browsable reference for all 661 scenario classes and 117 harm categories.

AILuminate Mapping Rationale

How we map to MLCommons AILuminate v1.0 taxonomy.

Evaluation

Grader Comparison Guide

Technical guide on automated grading tiers (Heuristic vs. LLM).

Grader Comparison Report

Technical analysis of grading reliability across 19,000 results.

Quick Links