dev:dry-run

To smoke-test an evolve pipeline, run a dry-run validation that checks Python tool syntax, argparse behavior, and end-to-end execution of setup, evaluation, and result tracing to catch integration issues safely before full experiments.

Offline pipeline validation checks Python syntax, argparse help output, and skill-to-tool cross-references to catch integration issues without requiring external services or API connections.

You need a LANGSMITH_API_KEY to run an end-to-end mock evaluation. When present, it triggers a mock agent through setup, run_eval, read_results, and trace_insights steps to validate the online pipeline.

You can validate argparse behavior and tool syntax in CI by running a dry-run smoke test that uses EVOLVER_TOOLS and EVOLVER_PY environment variables to locate and execute Python-based tool scripts for automated validation.

You should run a dry-run smoke test after installing or changing tools to ensure setup.py, run_eval.py, read_results.py, and trace_insights.py behave as expected and produce readable outputs before executing full optimization runs.