Core Trail Labs
Core Trail Kit
Terminology Doctrine
CTK introduces a deterministic operational vocabulary for understanding runtime behavior, contradictions, confidence shifts, and replayable decisions.
Most tools expose telemetry. CTK explains how runtime evidence becomes operational truth.
How CTK thinks
Step 1: Signals
Short explanation: Raw runtime observations from logs, health, status, topology, and incidents.
Operational meaning: Signals are raw input; they are never treated as truth alone.
Relation to next stage: Signals are normalized into evidentiary records.
Step 2: Evidence
Short explanation: Signals with source, scope, and freshness context recorded in the ledger.
Operational meaning: Evidence is audit-ready and replay-safe.
Relation to next stage: Evidence correlation can surface contradictions.
Step 3: Contradictions
Short explanation: Conflicting evidence states that cannot be trusted equally.
Operational meaning: Contradictions are risk multipliers, not noise.
Relation to next stage: Contradictions directly penalize confidence.
Step 4: Confidence Evolution
Short explanation: Trust score movement based on freshness, corroboration, and conflict pressure.
Operational meaning: Confidence tracks if decisions are becoming safer or riskier over time.
Relation to next stage: Current confidence feeds truth adjudication.
Step 5: Truth Adjudication
Short explanation: Deterministic resolution of current operational posture from evidence + confidence.
Operational meaning: One explainable runtime truth per evidence state.
Relation to next stage: Adjudicated truth is evaluated by safety policy.
Step 6: Safety Gate
Short explanation: Policy layer that allows, reviews, or blocks actions by trust posture.
Operational meaning: Prevents high-impact actions under weak evidence.
Relation to next stage: Every decision becomes replay lineage.
Step 7: Replay Chain
Short explanation: Deterministic chain from signals to final decision outcome.
Operational meaning: Operators can re-open why a decision happened.
Relation to next stage: Replay outputs become shared memory artifacts.
Step 8: Operational Memory
Short explanation: Persisted timeline of evidence, contradictions, confidence, and actions.
Operational meaning: Shift handoffs continue from truth, not recollection.
Relation to next stage: Memory seeds faster future adjudication.
Terminology cards
Operational Truth
Definition: The current runtime state CTK can defend with explicit evidence.
Why dashboards fail: Dashboards show slices; they rarely emit one defended conclusion.
How CTK handles it: CTK adjudicates one state from evidence quality, contradiction pressure, and confidence.
Operational meaning: You act on a reasoned state, not competing widget opinions.
Real example: Probe is green, but lag and timeout spikes produce degraded truth.
Related desktop screens: Timeline • Replay • Recommendations
Related concepts: Truth Adjudication • Confidence Evolution • Runtime Truth
Evidence Ledger
Definition: Immutable runtime evidence log with source, timestamp, and scope context.
Why dashboards fail: Signal history is often fragmented and hard to audit across tools.
How CTK handles it: CTK stores decision-linked evidence that is replayable and inspectable.
Operational meaning: Every recommendation can be traced to concrete evidence.
Real example: Blocked restart references stale health evidence and fresh runtime failure evidence.
Related desktop screens: Timeline • Inference • Incidents
Related concepts: Evidence Lineage • Operational Memory
Evidence Freshness
Definition: Recency quality of evidence used in decisioning.
Why dashboards fail: Old data can look valid and silently pollute live decisions.
How CTK handles it: CTK decays confidence when critical evidence streams become stale.
Operational meaning: Stale signals reduce action trust even if status appears stable.
Real example: No new logs for 8 minutes during pressure keeps restart blocked.
Related desktop screens: Signals • Recommendations • Timeline
Related concepts: Confidence • Safety Gate
Contradiction
Definition: A state where observed runtime evidence conflicts with expected operational posture.
Why dashboards fail: Isolated panels treat each signal family as independently true.
How CTK handles it: CTK creates explicit contradiction objects and propagates them into trust and policy decisions.
Operational meaning: Healthy checks can coexist with runtime failure risk.
Real example: Kafka lag rises while health endpoint remains green.
Related desktop screens: Consistency • Timeline • Recommendations • Signals
Related concepts: Confidence • Safety Gate • Replay Chain
Confidence
Definition: Current trust score for CTK's operational truth.
Why dashboards fail: Severity labels don't show whether decision trust is strong or weak.
How CTK handles it: CTK computes confidence from evidence quality, freshness, and contradiction pressure.
Operational meaning: Confidence determines whether actions should proceed or be gated.
Real example: Confidence drops from 0.82 to 0.41 after unresolved contradiction persists.
Related desktop screens: Insights • Inference • Recommendations
Related concepts: Confidence Evolution • Decision Trust Layer
Confidence Evolution
Definition: The directional movement of trust over incident time.
Why dashboards fail: Most systems show point-in-time labels, not trust transitions.
How CTK handles it: CTK persists confidence trajectory with causal evidence references.
Operational meaning: Operators see whether they are converging toward safe action.
Real example: Confidence recovers only after contradiction clears and fresh logs arrive.
Related desktop screens: Timeline • Inference • Replay
Related concepts: Confidence • Truth Adjudication
Truth Adjudication
Definition: Deterministic resolution step producing final runtime posture.
Why dashboards fail: Operators manually arbitrate conflicting evidence under pressure.
How CTK handles it: CTK applies explicit adjudication logic before recommendation emission.
Operational meaning: The team gets one explainable state to align around.
Real example: Service marked degraded despite green probe due to behavior contradiction.
Related desktop screens: Inference • Recommendations • Replay
Related concepts: Operational Truth • Safety Gate
Safety Gate
Definition: Policy control deciding whether action is allowed, review-only, or blocked.
Why dashboards fail: Action suggestions are often disconnected from trust state.
How CTK handles it: CTK gates action execution with confidence and contradiction-aware rules.
Operational meaning: Unsafe operations are explainably prevented.
Real example: Restart blocked while stale evidence and contradiction remain active.
Related desktop screens: Recommendations • Timeline • Incidents
Related concepts: Confidence • Decision Trust Layer
Replay Chain
Definition: Causal sequence from evidence intake to final decision.
Why dashboards fail: Historical context is fragmented across alerts and chat threads.
How CTK handles it: CTK stores deterministic decision branches and transitions for re-evaluation.
Operational meaning: Postmortems become reproducible instead of narrative-only.
Real example: Operator replays why scaling was allowed but restart stayed blocked.
Related desktop screens: Timeline • Replay Demos • Inference
Related concepts: Evidence Lineage • Operational Memory
Operational Memory
Definition: Shared, replayable memory of operational decisions and evidence states.
Why dashboards fail: Shift handoffs lose rationale and confidence context.
How CTK handles it: CTK timeline preserves evidence, contradiction, and decision lineage.
Operational meaning: Teams continue incident analysis without context reset.
Real example: Night shift replays day shift contradiction before remediation.
Related desktop screens: Timeline • Incidents • Walkthroughs
Related concepts: Replay Chain • Evidence Ledger
Incident Pressure
Definition: Composite urgency signal combining severity, trust, and contradiction weight.
Why dashboards fail: Alert counts ignore reliability of underlying evidence.
How CTK handles it: CTK ranks pressure by confidence state and contradiction context.
Operational meaning: Prioritization reflects decision risk, not only alert volume.
Real example: Medium severity with low confidence can outrank stable high severity.
Related desktop screens: Incidents • Insights • Recommendations
Related concepts: Confidence • Runtime Truth
Topology Delta
Definition: Confidence-aware change in service/entity relationships over time.
Why dashboards fail: Static topology snapshots hide drift relevance.
How CTK handles it: CTK tracks edge-level changes and maps them into incident reasoning.
Operational meaning: Dependency drift becomes actionable risk signal.
Real example: Consumer relation removed after deploy and confidence collapses.
Related desktop screens: Topology Tree • Relationship Inspector • Inference
Related concepts: Evidence Lineage • Contradiction
Decision Trust Layer
Definition: Combined action trust model derived from evidence, conflict, and policy.
Why dashboards fail: Recommendations lack explicit trust posture and risk gating.
How CTK handles it: CTK computes trust state and binds every recommendation to it.
Operational meaning: Operators know whether to execute, review, or pause.
Real example: Scale recommendation allowed; restart recommendation blocked in same incident.
Related desktop screens: Recommendations • Timeline • Inference
Related concepts: Safety Gate • Confidence Evolution
Runtime Truth
Definition: Live behavior reality established by current trustworthy evidence.
Why dashboards fail: Status panels can lag or overfit one signal source.
How CTK handles it: CTK continuously re-adjudicates truth as evidence evolves.
Operational meaning: Operators can pivot with state changes confidently.
Real example: Runtime truth remains degraded until lag and timeout curves recover.
Related desktop screens: Signals • Insights • Recommendations
Related concepts: Operational Truth • Confidence
Evidence Lineage
Definition: Traceable path connecting evidence records to decisions.
Why dashboards fail: Most tools cannot prove the lineage behind operational advice.
How CTK handles it: CTK links evidence IDs, contradiction states, and confidence transitions in replay chain.
Operational meaning: Audits and postmortems remain defensible under scrutiny.
Real example: A blocked action points to exact evidence set and failed gate threshold.
Related desktop screens: Replay • Timeline • Recommendations
Related concepts: Evidence Ledger • Replay Chain
What CTK is not
CTK is not
CTK focuses on
Docs to Trial Path
1. Replay first
Open a deterministic replay to see evidence, contradiction, confidence, and gate state in one chain.
Open Replay Demo2. Install desktop
Download CTK Desktop and run local-first observation for your workspace runtime boundaries.
Download Desktop3. Initialize workspace
Connect repo + source mode, then start deterministic timeline and contradiction-aware recommendations.
Initialize Workspace