Pre-release readiness

Recipe — pre-release readiness evaluation gate.

Run a curated regression evaluation before any production release. Block the release on regression.

from layerlens import Stratix
client = Stratix()

# Run your release-gating evaluation (model + benchmark you've designated for releases)
model = client.models.get_by_key(os.environ["RELEASE_MODEL_KEY"])
benchmark = client.benchmarks.get_by_key(os.environ["RELEASE_BENCHMARK_KEY"])
evaluation = client.evaluations.create(model=model, benchmark=benchmark)
evaluation = client.evaluations.wait_for_completion(evaluation, timeout_seconds=1800)

assert evaluation.is_success, f"Release-gate eval failed: status={evaluation.status}"
assert evaluation.accuracy >= 0.95, f"Release-gate eval below threshold: accuracy={evaluation.accuracy}"

Wire from your release pipeline (CI, deploy script, manual checklist tool). Fail-loud if either condition is false.

See also

• Workflow: Govern
• Tutorial 3: CI/CD gates