Model Compatibility Smoke Test Pack 20260831 v1.0
What it does
Model Compatibility Smoke Test Pack helps multi-model agent teams review whether a skill has enough evidence to behave consistently across models. It checks redacted local notes for model matrix, prompt fixtures, tool-call results, structured-output results, timeout behavior, fallback results, deterministic comparison, and review ownership.
Use cases
- Platform teams validating model-routing changes before production use.
- AI operations teams comparing whether a skill works across current and candidate models.
- Reviewers checking smoke-test evidence before approving a model onboarding package.
Requirements
- OpenClaw skill runtime with local file access to the skill directory.
- Python 3 standard library only; no third-party dependencies.
- No model API keys required. The skill reviews existing redacted evidence and does not call model endpoints.
Example usage
python3 scripts/model_compatibility_smoke_test_pack.py --notes path/to/redacted-notes.txt --json
Expected output: JSON showing PASS, REVIEW, or BLOCKED, with evidence coverage, risks, recommendations, deterministic comparison findings, and mutated_files=[].
Price: $4.99