eval-harness (nano-step)

Behavior-regression testing harness for LLM agents with cost gating and flaky detection.

Servicefreeglobal

eval-harness is a free, typed evaluation harness for testing LLM agent skills for behavioral regressions, with dollar-cost gating and flaky-result detection, built on bash and jq and working with OpenCode today with pluggable support for other runners. It's aimed at developers building LLM agent skills who need to catch regressions in behavior or cost automatically rather than discovering them after the fact.

Website

Full match profile

Behind the summary, Matchbox keeps a richer profile of eval-harness (nano-step) - the signals our matcher actually reads to decide when to surface it. It stays private; claim the listing to see and control it.

Problem & pain-point mapping
Who we surface it to (audience fit)
What it's a strong alternative to
Trust & credibility signals

Claim eval-harness (nano-step) to unlock →

Something wrong with this listing — dead link, not a real product, wrong info?

Try Matchbox with your own problem

Describe what is not working - we’ll show you whether eval-harness (nano-step) (or something else) actually fits.

Try a problem