momentohq/mo-eval
v0.3.0
Workflows for benchmarking agents and preparing coding tasks with mo-eval.
author-task
Use when a request needs an unlisted mo-eval task draft packaged from a fixed Git base; not for publishing a task or running an evaluation.
compare
Use when a request needs a controlled comparison across multiple configurations to design, run, or interpret; not for a baseline for one configuration (use evaluate).
evaluate
Use when a request needs a baseline for one configuration to run, read, or iterate; not for a controlled comparison across multiple configurations (use compare).
prepare-repository
Use when a request supplies a task runtime or dependency preparation to record; not for inferring a repository runtime or authoring a task.