Reference

Topics

6 topics

Six topics, one page each. A topic page gathers the research, guides, terms, and calculators filed under it, wherever on the site they live.

Multi-agent failures 21 pieces How systems of cooperating agents break: coordination failures, role confusion, and the taxonomies that name them.
Propagation & containment 18 pieces How a single bad step spreads through a pipeline, and what actually contains the propagation radius.
Reliability testing 20 pieces Testing agents like software: harnesses, regression suites, and what a passing run does and does not prove.
Eval statistics 27 pieces The statistics under an eval result: sample size, confidence intervals, the tests that separate a real difference from noise, and what an eval can certify.
Judge reliability 16 pieces Using a model to score another model: bias, position effects, agreement with humans, and how far a judge can be trusted.
RAG & retrieval 13 pieces Retrieval as a reliability problem: chunking, faithfulness, and the failure modes that live between the index and the answer.