Answer
Question, evidence, uncertainty, proof debt, next evidence
Glass LabProof-first, layered detail
Inspect what KELLY knows, what it does not know, and why.
Glass Lab is the transparency console. It turns answers, forecasts, scenarios, settlements, and memory into inspectable objects without overwhelming first-time users.
Loop degraded
Proof Pulse
Live checks show memory birth resumed, but the continuous loop is still degraded. This panel refuses to imply a passing runtime.
Needs closure
Live Loop Watch
Tracks the next required proof: a clean production memory-loop cycle with activation event publication and heartbeat closure.
Verified receipts required
Active Fleet
Workers appear active only when fresh receipts prove current activity. No fake animations, no decorative agent motion.
Inspectable
Proof Debt
Shows what evidence is missing: settlement, calibration, replay, freshness, sample size, or independent reproduction.
Gate planned
Thought Stream
KELLY may publish governed cognitive readouts, not raw private thoughts. Every item needs proof status and interruption policy.
Memory birth resumed
Memory Fabric
New governed memory cells exist again; each cell still needs clear birth chain, ancestry, proof state, and invalidation rules.
Operator visible
Repair Desk
Blocked, degraded, or broken systems should become diagnostics and recovery work orders, not silent failure.
Inspectable objects
One trail for every serious output.
Forecast
Lock, horizon, probability, resolver, settlement plan
Scenario run
Inputs, model reference, synthetic label, replay class
Settlement
Outcome, resolver receipt, score, calibration link
Memory cell
Birth chain, ancestry, proof state, retrieval posture
Proof packet
Exportable evidence bundle for reviewers
Teach through use
Probability, causality, uncertainty, and proof become understandable through guided inspection, not jargon-first screens.
Separate synthetic from empirical
Scenario runs can be useful, but Glass Lab keeps them below reality-settled evidence and governed memory.
Show lineage
Reviewers should be able to trace what was asked, what was retrieved, what was scored, and what changed later.