Look inside AI systems.

Notes on AI agents, language models, and software systems—grounded in code, traces, and reproducible experiments.

All writing Agent Runtime Lab Archive

AI AGENT · LANGUAGE MODEL · SOFTWARE SYSTEMS

Three ways in

Choose a path by the question you are trying to answer. The archive does not need to be read in order.

All writing →

Authorization decides whether one principal may perform one action on one resource under specific conditions

Concurrency shortens only independent waiting time and adds result joining, cancellation, and resource pressure

Longer retention does not create stronger knowledge; unverified memory carries old errors into new tasks

A plausible answer does not prove that evidence is real, execution was authorized, or interruption recovery is correct

The model does not receive one prompt. It receives context blocks assembled from different sources for different purposes.

The model does not observe local execution; the runtime must carry evidence into the next turn

The model does not operate the system directly; it selects from the actions exposed by the runtime

Retrying a model call spends computation; retrying a send, payment, or create operation can change the external world twice