Quality Gate

Graduated pass/fail evaluation with four ordered levels. Caller defines the mapping from observed metrics (confidence scores, test counts, signal strength) to a QualityLevel; the gate stays opinion-free.

Quality levels

Level	Numeric	Use
MINIMAL	1	Basic correctness — runs without crashing
STANDARD	2	Acceptable for an internal handoff
STRICT	3	Reviewable; ready for further work
RELEASE	4	Production-grade

These names are advisory; what matters is the ordering. QualityLevel is an IntEnum so >= works.

Usage

from techrevati.runtime import QualityGate, QualityLevel

gate = QualityGate(required_level=QualityLevel.STRICT)
observed = my_quality_estimator(...)   # caller-defined mapping → QualityLevel

outcome = gate.evaluate(observed)
if outcome.satisfied:
    advance()
else:
    request_rework(outcome)

API

class QualityLevel(IntEnum): MINIMAL=1, STANDARD=2, STRICT=3, RELEASE=4

@dataclass(frozen=True)
class QualityGate:
    required_level: QualityLevel
    def evaluate(self, observed: QualityLevel | None) -> QualityGateOutcome: ...
    def is_satisfied_by(self, observed: QualityLevel) -> bool: ...

@dataclass(frozen=True)
class QualityGateOutcome:
    satisfied: bool
    required_level: QualityLevel
    observed_level: QualityLevel | None

None as observed always fails — no signal means no gate satisfaction.