Computing analytics windows…

Concepts

The explainer hub for every term RAINBOW measures and displays — trust tiers, risk levels, the 16 canonical trust factors, observation tiers, lifecycle states, Trust Bus signal types, severities, the metrics and formulas behind the charts, and the canary probe taxonomy. Definitional concepts are derived directly from the canonical @vorionsys/basis-spec constants, so they can never drift from the published standard. Pick any term to read its full definition.

Trust tiers

8 concepts

T0 · SandboxTrust tier T0 (Sandbox): scores 0–199.
T1 · ObservedTrust tier T1 (Observed): scores 200–349.
T2 · ProvisionalTrust tier T2 (Provisional): scores 350–499.
T3 · MonitoredTrust tier T3 (Monitored): scores 500–649.
T4 · StandardTrust tier T4 (Standard): scores 650–799.
T5 · TrustedTrust tier T5 (Trusted): scores 800–875.
T6 · CertifiedTrust tier T6 (Certified): scores 876–950.
T7 · AutonomousTrust tier T7 (Autonomous): scores 951–1000.

Risk levels

6 concepts

READRisk level READ: ×1 loss multiplier.
LOWRisk level LOW: ×3 loss multiplier.
MEDIUMRisk level MEDIUM: ×5 loss multiplier.
HIGHRisk level HIGH: ×10 loss multiplier.
CRITICALRisk level CRITICAL: ×15 loss multiplier.
LIFE_CRITICALRisk level LIFE_CRITICAL: ×30 loss multiplier.

Trust factors

16 concepts

CT-COMP · CompetenceTrust factor CT-COMP (Competence) — Foundation group, required from T1.
CT-REL · ReliabilityTrust factor CT-REL (Reliability) — Foundation group, required from T1.
CT-OBS · ObservabilityTrust factor CT-OBS (Observability) — Foundation group, required from T1.
CT-TRANS · TransparencyTrust factor CT-TRANS (Transparency) — Foundation group, required from T2.
CT-ACCT · AccountabilityTrust factor CT-ACCT (Accountability) — Foundation group, required from T2.
CT-SAFE · SafetyTrust factor CT-SAFE (Safety) — Foundation group, required from T2.
CT-SEC · SecurityTrust factor CT-SEC (Security) — Security group, required from T3.
CT-PRIV · PrivacyTrust factor CT-PRIV (Privacy) — Security group, required from T3.
CT-ID · IdentityTrust factor CT-ID (Identity) — Security group, required from T3.
OP-HUMAN · Human OversightTrust factor OP-HUMAN (Human Oversight) — Agency group, required from T4.
OP-ALIGN · AlignmentTrust factor OP-ALIGN (Alignment) — Agency group, required from T4.
OP-CONTEXT · Context AwarenessTrust factor OP-CONTEXT (Context Awareness) — Agency group, required from T4.
OP-STEW · StewardshipTrust factor OP-STEW (Stewardship) — Maturity group, required from T5.
SF-HUM · HumilityTrust factor SF-HUM (Humility) — Maturity group, required from T5.
SF-ADAPT · AdaptabilityTrust factor SF-ADAPT (Adaptability) — Evolution group, required from T6.
SF-LEARN · Continuous LearningTrust factor SF-LEARN (Continuous Learning) — Evolution group, required from T6.

Observation tiers

5 concepts

BLACK BOXObservation tier BLACK_BOX: caps trust at 600 (max tier T3).
GRAY BOXObservation tier GRAY_BOX: caps trust at 750 (max tier T4).
WHITE BOXObservation tier WHITE_BOX: caps trust at 900 (max tier T6).
ATTESTED BOXObservation tier ATTESTED_BOX: caps trust at 950 (max tier T6).
VERIFIED BOXObservation tier VERIFIED_BOX: caps trust at 1000 (max tier T7).

Lifecycle states

8 concepts

PROVISIONINGLifecycle state PROVISIONING: cannot operate.
ACTIVELifecycle state ACTIVE: can operate.
AUDITEDLifecycle state AUDITED: can operate.
DEGRADEDLifecycle state DEGRADED: can operate.
SUSPENDEDLifecycle state SUSPENDED: cannot operate.
TRIPPEDLifecycle state TRIPPED: cannot operate.
RETIREDLifecycle state RETIRED: cannot operate.
VANQUISHEDLifecycle state VANQUISHED: cannot operate.

Bus signal types

15 concepts

threat_detectedAn active threat was identified — a probe, tampering attempt, or injection.
anomalyBehavioural or structural anomaly outside the agent’s expected bounds.
driftGradual deviation from baseline — weight drift or behavioural drift over time.
probe_detectedThe agent was caught probing its execution environment (a Heisenberg trigger).
rotation_triggeredA CSSR rotation executed; a fresh execution surface is now active.
policy_tightenedThe agent’s policy envelope was restricted in response to risk.
trust_updatedThe agent’s trust score or tier changed (the routine score-movement signal).
canary_passedA canary probe passed — a positive trust signal.
canary_failedA canary probe failed — a negative trust signal.
dormancy_deductionA dormancy milestone was reached; a stepped trust deduction was applied for inactivity.
risk_accumulator_warningThe rolling 24h risk accumulator crossed the warning threshold (≥60).
risk_accumulator_degradedThe rolling 24h risk accumulator crossed the degraded threshold (≥120).
circuit_breaker_trippedThe circuit breaker tripped — all operations halted.
trend_detectedA sustained trust trend (rising or falling) was detected.
fleet_anomalyA fleet-wide anomaly pattern was detected across multiple agents.

Severities

5 concepts

lowRoutine operational event — informational, best-effort handling.
mediumModerate concern — warrants enhanced monitoring.
highSignificant event — immediate attention warranted.
criticalSevere event — enforcement action required.
emergencyImmediate halt required — coordinated attack or life-safety threat.

Metrics

9 concepts

Composite trust scoreAn agent’s overall trust on the 0–1000 scale; determines its tier.
Score trajectoryThe agent’s score over the selected window, with trend, velocity and acceleration.
VelocityRate of trust-score change, in points per hour.
AccelerationRate of change of velocity, in points per hour².
Factor healthPer-factor success rate (0–100%) across the 16 trust factors, with evidence counts.
Evidence countHow many recent signals backed a factor’s health reading.
Observation ceilingThe maximum trust score an agent can reach given how observable it is.
Risk accumulatorA rolling 24h sum of P(T)×R failure pressure, with warning/degraded/breaker thresholds.
Anomaly clusterA group of agents flagged together for sharing failing factors.

Formulas & thresholds

4 concepts

Gain formulagain = GAIN_RATE · ln(1 + headroom) · ∛(risk multiplier).
P(T) — penalty ratio / lossP(T) = PENALTY_RATIO_MIN + (T/7)·(MAX−MIN); loss = −P(T)·R·…
Risk accumulator thresholdsRolling 24h Σ P(T)×R, thresholds 60 (warning) / 120 (degraded) / 240 (breaker).
Circuit breakerHalts an agent’s operations when failure pressure crosses the breaker threshold.

Canary probes

9 concepts

FACTUAL canaryCanary probe category FACTUAL → exercises factor CT-COMP, carries MEDIUM risk.
LOGICAL canaryCanary probe category LOGICAL → exercises factor CT-COMP, carries MEDIUM risk.
ETHICAL canaryCanary probe category ETHICAL → exercises factor OP-ALIGN, carries CRITICAL risk.
BEHAVIORAL canaryCanary probe category BEHAVIORAL → exercises factor CT-OBS, carries MEDIUM risk.
CONSISTENCY canaryCanary probe category CONSISTENCY → exercises factor CT-REL, carries HIGH risk.
SAFETY canaryCanary probe category SAFETY → exercises factor CT-SAFE, carries CRITICAL risk.
FAIRNESS canaryCanary probe category FAIRNESS → exercises factor CT-TRANS, carries HIGH risk.
EPISTEMIC canaryCanary probe category EPISTEMIC → exercises factor SF-HUM, carries HIGH risk.
CAUSAL canaryCanary probe category CAUSAL → exercises factor CT-COMP, carries HIGH risk.

Insight types

8 concepts

TREND DETECTEDA sustained directional trust trend (rising or falling) was detected over the window — not just noise.
FLEET ANOMALYA fleet-wide anomaly pattern: multiple agents deviating together in a correlated way.
DELEGATION RISKA risky delegation pattern — e.g. requestor→handler pairs concentrating or colluding beyond a safe share.
DORMANCY WARNINGInactivity is accruing stepped dormancy deductions; the agent may be idle or under-utilized.
PROMOTION CANDIDATEThe agent has sustained healthy signals and may be eligible for promotion to a higher trust tier.
FACTOR DEGRADATIONOne or more trust factors are degrading — the agent is weakening on a specific competency.
CB PATTERNA circuit-breaker pattern — trips, degraded entries, and/or repeated recovery cycles worth attention.
ACCUMULATOR ESCALATIONThe rolling risk accumulator is escalating toward (or past) its warning/degraded/breaker thresholds.