Technical experiments, articles, courses, and labs on coding agents, coordination, evaluation, and software architecture.
Measuring agent count against time and token count to code, merge, and test against one repository.
Read the experiment →Inspect the methods, look at the evidence, and understand the limitations of technical experiments.
View experiments →Follow structured pathways through coding agents, coordination, and software architecture.
View courses →Work through interactive simulations and guided exercises.
View labs →A short introduction to agents working at the same time, why it can help, and where coordination adds work.
Measuring agent count against time and token count to code, merge, and test against one repository.
How do agent concurrency and a persistent full-workload strategy affect completion time and token usage for ten tested coding tasks?
See how concurrency, shared files, integration queues, and task batching affect the time it takes to get ten changes merged and verified.
See how concurrency, shared files, integration queues, and task batching affect the time it takes to get ten changes merged and verified.
Experiments document their workloads, methods, results, and limitations.
Published findings are distinguished from interpretations. Reproducibility status and supporting artifacts are included with each experiment.
About the project →Follow new publications with RSS