Issue #673 · Production parity verification

Questionnaire Thinking Parity

This report compares Claude.ai and production Tegy for the same colors questionnaire prompt. The old second keyframe gap was a visible Thinking label in Tegy; production now shows the working mark only.

What Changed

Active assistant turns preserve the accessible status Tegy is thinking, but remove visible status text.

Questionnaire-ready turns keep the Claude-like sequence: message actions, working mark, questionnaire card, direct reply composer.

Production verification found no visible live-status text, no dead Thought for Ns row, one questionnaire, and one direct composer.

Keyframes

Claude in-flight reference with user bubble and working mark — Claude.ai second keyframe reference.

Tegy production in-flight state with user bubble and working mark only — Tegy production second keyframe after deploy: no visible `Thinking` label.

Claude ready questionnaire card with direct reply composer — Claude.ai ready questionnaire reference.

Tegy production ready questionnaire card with direct reply composer — Tegy production ready state after deploy. The model included short option descriptions; these are assistant/tool output, not UI chrome.

Verification

Production chat used for final capture: app.tegy.io/chat/451e74fd-aee5-4386-8f06-4215cb73ccc0.

Tested commands included focused browser regressions for questionnaire rendering, active thinking state, composer state, and npm run build. Production was deployed with Worker version 237072b8-a4a3-4e0a-a5c7-03cff55cee4a.