Agent Readability Lab

The same moment. One feed an agent can reason with.

Every agent reads markets through flat numbers β€” price, volume, RSI. SupraScope turns the same moment into a structural narrative: memory layers, layer divergence, deviation fingerprints, topology events. Below, the same market moment is handed to an agent both ways, with the same question. A neutral referee β€” told nothing about which feed is which β€” judges whose reasoning is sharper.

Pick a market moment
The same moment, described two ways
Raw feedwhat agents get today
Structural narrativeSupraScope API output
Ask both the same question
Generates fresh responses + a blind referee verdict. Runs live inside Claude; on the public site it shows the example responses below.
Two agents, same moment
Agent fed the raw feedExample
Agent fed the structural narrativeExample
Neutral referee β€” which briefing produced better reasoning
What you're looking at. The structural panel is generated by narrate(state) β€” the same transform that powers the SupraScope agent API, run on a row from the live states table. The example agent responses are representative; Run it live generates fresh ones and a blind referee verdict in real time. The raw-feed responses aren't strawmen β€” they're a capable agent doing its honest best with the only data it has. The structural agent reasons further because the briefing carries context the flat numbers structurally cannot.