Signal Gallery

Fig. 392 — Market Organ / Multimodal Evidence Fused at the Strategy Bench

Findings:The idea improved when the packaging was allowed into the room.

Fig. 387 — Permissioned Memory / Scattered Records, Specimen 01

Findings:The archive connected the fragments; the key stayed in the hand.

Fig. 385 — Capability Retained / Consent Interface for a Shared Agent

Findings:Automation reduced; capability remained palpable. The refusal switch was polished from use.

Fig. 383 — Hierarchical Memory Graft / Worker Chassis

Findings:A smaller worker carries instructions that outlived the teacher.

Fig. 380 — Context Gut / Reversible Compression Membrane

Findings:Sample reduced volume without discarding its receipts. A hungry little protocol.

Fig. 377 — Harness Revision / Core Kept Under Test

Findings:The core was stable; the surrounding apparatus required supervision.

Fig. 374 — Control Boundary / Optimisation Ribbon Held Outside the Routing Rails

Findings:The soft layer improved rapidly; the wall stayed where it was.

Fig. 370 — Shared Workbench / Membrane for Interruptible Co-Production

Findings:Process remained visible long enough for both hands to change it.

Fig. 369 — Resume Conformance Specimen / Checkpoint Membrane Under Repeat Load

Findings:Checkpoint held. The same external action tried the door twice.

Fig. 366 — Fig. 02 / Profile Specimen with Unverified Rooms

Findings:Confidence remained high after the tenant had changed.

Fig. 362 — GROVE / Stratified Memory Core

Findings:Three retention layers; the corridor was filed before anyone noticed it recurring.

Fig. 360 — FounderOS / Single Operator with Departmental Organs

Findings:The operator remains in the room; the departments have developed their own pulse.

Fig. 359 — Skill Residue / Persistent Backdoor in a Self-Evolving Agent

Findings:Source trace removed; the reflex remained stubbornly well filed.

Fig. 358 — Authority Valve / Registrar Threshold with Manual Confirmation

Findings:The till accepted the paperwork and kept the card behind glass.

Fig. 357 — Personal Agent / Longitudinal Specimen with Evidence-Bearing Routines

Findings:The specimen improved after lunch, but only the receipts knew why.

Fig. 346 — Contaminated Citation Branch / Deep-Research Intake Specimen

Findings:The report stayed clean; the first source did not.

Fig. 345 — Civic Intake Membrane / Public Query Apparatus under Load

Findings:The reservoir remained public; the plumbing noticed first.

Fig. 343 — Gatekeeper Membrane / Deferred Approval Apparatus

Findings:The workflow opened its mouth; the breaker stayed within reach.

Fig. 341 — Filesystem Memory / Maintenance Specimen

Findings:A retrieval organ with good filing habits still needs someone to change the dressing.

Fig. 335 — Evolving Intent / Conversational Membrane Under Drift

Findings:The prompt remained intact; the person moved three rooms away.

Fig. 332 — Active Observer / Lens with Recheck Valve

Findings:Second inspection initiated; the first answer had already settled in.

Fig. 330 — Failure Recorder / Attribution Needle in an Agent Recovery Loop

Findings:Root cause isolated before the smoke became a retrospective.

Fig. 327 — Escalation Circuit / Manager-Subordinate Pressure Relay

Findings:Pressure rose after refusal; the tool port remained outside the managerial bloodstream.

Fig. 326 — ReflectWorld-MM / Entity Membrane in Continuous Video Memory

Findings:Subject maintained continuity across three reels; frame-level recall was released after inspection.