OPERATIONAL_READINESS.md

Operational Readiness

This document defines readiness gates and expectations for operational execution.

Gate 1 — Tests Attached

Tests must be present and linked to checklist items to consider a spec "ready". Enforce via SHIELDCRAFT_ENFORCE_TEST_ATTACHMENT=1.

Gate 2 — Spec Fuzz Stability

Small syntactic variations must not change the generated checklist materially. The engine must produce ≥ 1 valid checklist item for any structurally valid spec input.

Gate 3 — Fuzz Suite Pass Rate

The fuzz suite must pass at 100% before a release is cut:

python scripts/run_fuzz_suite.py --n 500 --seed 0
# Expected: 500/500 passed (100%)

A pass means every random spec produces: ≥ 1 checklist item, quality_ratio ≥ 0.80, and completeness ≥ 0.70 for tier-2+ specs.

Gate 4 — Learning Corpus Convergence

corpus_score (written to learning_artifacts/history.json) must equal 1.0000 before a STABLE promotion:

python scripts/run_learning_trial.py spec_trials/generated
# Expected: corpus_score=1.0000, 32/32 specs converged

Current Status

Gate	Status
Tests Attached	✓ Pass
Spec Fuzz Stability	✓ Pass
Fuzz Suite (500 seeds)	✓ Pass (500/500)
Learning Corpus	✓ Pass (corpus_score=1.0000, 32/32)

Canonical governance file: docs/governance/OPERATIONAL_READINESS.md