# Blender Headless Automation Lab

You receive eight synthetic job records. Your task is to implement a small evaluator in `starter.py` that keeps job qualification, process execution, and artifact validation separate.

Required result fields:

- `version_pinned`
- `inputs_bound`
- `cli_order_valid`
- `trust_gate`
- `process_state`
- `artifact_state`
- `final_decision`
- `handoff_ready`

Do not execute Blender or call an external service. The fixtures are local and synthetic. Use `run_checks.py` first to verify that the learner-visible fixture set is complete. Your evaluator should fail closed: missing trust, missing assets, invalid argument order, timeout, process failure, or invalid output must never become `VERIFIED_RESULT`.