Analyzer, Intent Builder, and Workflow UI overhaul

Version: 2.3.1, Date: 2026-07-21

A UI overhaul lands across the Analyzer, Intent Builder, and Workflow surfaces, alongside three functional additions: AI-generated short descriptions, SEO-friendly generated prompt titles, and a compatibility fix for newer Claude models. The rest of the post is grouped by app area.

Highlights

• Unified card design across the app: the Analyzer results, Intent Builder results, Create Analyzer, and My Analyzers pages now share one consistent card pattern, so moving between them feels the same.
• Real analyzer status in the Analyses list: rounds now show their actual state (Completed, Incomplete, or in progress) instead of always reading “Completed.”
• AI-generated short descriptions: the Short Description field on an analyzer now has a Generate button that produces a clean, end-user-facing one-liner.
• SEO-friendly prompt titles: generated prompts in the Intent Builder now come with titles that follow SEO best practices out of the box.

Prompt Analyzers

• Improved: The Prompt Input card on the analyzer run page has been recolored and the Model tag has been moved onto the title row and right-aligned, so the header is tighter and the model in use is easier to spot.
• Improved: Improve, Strengthen, Transform, Adapt, and Tune have all been recolored to green and their popup menus have been tightened: smaller row gaps, bullet markers, and no wrapping on desktop, so the choices are easier to scan.
• Fixed: The Analyses list on the run page now reflects the real status of each round. Previously every round would show “Completed” even while insights or ratings were still working or had failed. It now derives Completed, Incomplete, or in progress from the actual state.
• New: The Short Description field on an analyzer now has an AI Generate button. It produces a concise, end-user-facing one-liner from the analyzer’s name and category, capped at 100 characters. The generated text is the same description that surfaces in the end-user help tooltip.
• Improved: The Create Analyzer and My Analyzers pages now show the Analyzer Content card and the Test Analyzer card side by side in an equal-height layout. Both pages carry the full Test Analyzer treatment: a “Sample Prompt” and “Test result” title row, a four-icon sample toolbar (expand, raw versus formatted, copy, export), a centered empty state, and an always-visible result card so it is clear where the output will appear.
• Improved: Card badges (Required and Optional), button colors, and corner radii on the Create and My Analyzers pages have been normalized to match the shared button format used elsewhere in the app.

Intent Builder

• Improved: The generated prompt result is now a single, self-contained card. Before, a floating header sat above a separately bordered content box, which read as a double border. Now one card carries the header (prompt title on the left, action icons at the upper right), a separator, and a scrolling body, matching the analyzer results pattern.
• Improved: The result card header now shows “Prompt X of Y” and a Completed badge, along with a right-aligned Model tag so the model that produced the result is always visible.
• Improved: The Generated Prompts column on the left has been restyled to match the shared card pattern, widened to 25% of the page, and de-chromed so it feels lighter. Long titles now wrap instead of pushing the action icons off screen.
• Improved: Regenerate has been moved to a labelled green button at the bottom of the card, Save buttons gray out when disabled, and Generate has been normalized to the standard button height, so the actions are easier to find and their state is clearer.
• Improved: The Prompt Intent Builder card and the results card have been recolored to give the two stages of the flow visually distinct identities.
• New: Generated prompt titles now follow SEO best practices out of the box: at most 60 characters, keyword-rich, and never a repeat of the badge path or the intent description. Applies to the single, combined, and unified generation flows for newly generated prompts.

Workflows

• Improved: The Pause, Rerun Step, and Rerun Workflow buttons are now always visible, colored green when available and gray when their conditions are not met. Previously Rerun Step would disappear entirely in some states, which made the workflow feel like it had lost the action.
• Improved: The step header now shows the AI model, right-aligned on the title row, so you can see at a glance which model each step is using.
• Fixed: The step title and description inputs now show a single teal border on focus. Before, the field displayed a purple focus ring plus a black outline, which read as two competing borders.
• Improved: The Save, Update, and Create buttons have been recolored to match the standardized button palette used across the app.

Across the App

• Improved: The model selector used throughout the app has been slimmed down. Each model row is now a single compact line, with the extra Quality percentage and context-tokens text removed and padding tightened, so choosing a model is quicker.
• Improved: The library dropdown used when saving to a library now shows a single violet focus border instead of a ring plus a border, matching the rest of the app’s focus treatment.
• Fixed: Newer Claude models (Sonnet 5, Opus 4.8, Haiku 4.5, and Fable 5) reject the temperature parameter and were returning errors. The app now sends temperature only to the models that support it, so runs against the newer Claude family complete cleanly. Older Claude 3.x models and other providers are unaffected.