siddhant-k-code/agent-trace eval

Run agent-trace evals in CI, post results to the step summary, and fail on regression.

View on GitHub

Trust Signals

Scorecard Score
not yet scored
Maintenance Recency
Stale
License
None
namedescriptionrequireddefault
configPath to eval config file.agent-evals.yaml
baselinePath to baseline scores file (optional)""
save-baselineSave scores as new baseline after runfalse
toleranceRegression tolerance (0.0–1.0, default 0.05)0.05
trace-dirDirectory where agent-trace sessions are stored.agent-traces
python-versionPython version to use3.12
install-extrasComma-separated optional extras to install (e.g. openai,anthropic)""
namedescription
passedtrue if all scorers passed, false otherwise
summary-pathPath to the written eval summary markdown file