validate

You can validate config.yaml and Python artifacts by running canonicalize, dry_fire, and contract validators to verify integrity before ending a round. This self-validation workflow checks configuration files, tool scripts, processors, and templates for correctness.

Synthetic replay validation is a deterministic check that tests artifact integrity by simulating execution paths. It operates alongside canonicalization, dry-fire, and contract checks to ensure tools, processors, and templates function correctly before workflow completion.

Yes, you need a Python environment with harnessx.meta_harness.validate_workflow modules to run self-validation checks on HarnessX artifacts. The validators require this specific Python setup to execute canonicalization, dry-fire, contract, and synthetic replay checks.

You can scan Python tools and processors for hardcoded literals using an advisory validator that emits non-blocking warnings and JSON reports. This literals scanning checks files in the tools and processors directories without halting the workflow.

The best way to enforce policy checks for novelty and evidence on non-noop rounds is through the post-flight validation workflow. These policy blocks run after end_turn alongside validity checks, ensuring artifacts meet evidence requirements without blocking advisory scans.