← Previous9 / 25Next →

Composal Deep Scan Runs

ComposalBrowser agents that test your real user flows and catch bugs before your users do

Design with AI

Create a new design inspired by Composal Deep Scan Runs, with your own content.

1Content

2Fidelity to the reference

Keeps its visual language, rearranges freely around your content.

3Your brand Optional

Use the reference style

Brand nameColorsFontsGuidelines

4Output

Prompt 752 words

# Design from Reference: Composal Deep Scan Runs

Create an original, high-fidelity visual design inspired by the reference. Prioritize composition, typography, colors, spacing, and component styling over technical implementation.

## Reference

- **Product:** Composal
- **Category:** Analytics
- **Screenshot:** attached to this message (also available at https://media.saasframe.io/v/9c893df8c37f5dee3777-1600.webp)
- **Source:** https://composal.com/saasframe/verify/acme-notes/deepscan

Analyze the attached screenshot first. If no image is attached, open the screenshot link; if neither is accessible, ask for it rather than guessing.

## Visual Design Context

## Overall aesthetic
Composal is a QA platform where browser agents test real user flows. The marketing site is clean and bright: white pages, very large tight headlines, product UI fragments (pull request checks, run timelines, terminal output) and a signature acid-yellow accent. The Verify app is a dense, developer-tool dark interface in near-black, close to Linear or Vercel.

## Typography
Geist everywhere, with Geist Mono for prompts, commands and code. Marketing headlines are Geist 600 at 60–96px with very tight tracking (about -3px at 84px); body copy 16–18px. The app uses small type: 18px semibold page titles, 13–14px labels, 11–12px metadata.

## Colors & surfaces
Marketing: #FFFFFF backgrounds, #F6F8FA and #F3F4F6 panels, near-black #111827 / #161616 text and UI mockups, acid yellow #F1F93C for highlights and primary buttons, rose #F43F5E for failures. App: #0A0A0A to #161616 surfaces, hairline borders, white primary buttons, yellow #F1F93C reserved for the demo badge and key calls to action, green/red/orange status dots.

## Components
Product UI fragments as illustrations (scenario runs on a pull request, failure cards, product maps), terminal blocks, numbered step lists, waitlist form with invite code. App: collapsible sidebar with project switcher, tables with priority pills (P0–P4) and status dots, run timelines with video recording and step logs, stat rows, empty states with a single action, onboarding cards with chevrons.

## Layout
Marketing content centered around 1200px with alternating text and UI fragment sections; generous white space. The app uses a 240px sidebar and a content column up to about 1480px; onboarding steps are a single centered 600px column with the logo top-left and "Exit setup" top-right.

## Tone
Direct, engineer-to-engineer copy in sentence case: "Find bugs before your users do", "What should we scan?", "Add a test user to start your scan."

### Overall aesthetic
Dark run history table.

### Page composition
Title with subline and action, filter bar, table of runs.

### Typography
Geist 18px title, 13px run names, 12px metadata.

### Colors & surfaces
Near-black page, findings pills, green Completed and orange Partial labels.

### Components
Segmented toggle, status select, pill buttons inside table cells.

### Responsive
Duration and start columns hide on small screens.

### Adaptation guidance
Put the outcome (findings, coverage) next to each run's name so users can scan for the ones that matter.

## Content

**Instructions:**

Generate realistic example content that fits the selected design reference.

Create appropriate headlines, descriptions, labels, buttons, and supporting copy based on the type of interface being designed.

If a product or business description is provided, use it to guide the content.

Otherwise, invent a plausible fictional SaaS product appropriate for the reference.

Keep the content concise, coherent, and visually balanced.

Do not present fictional statistics, customer names, or testimonials as verified facts.

## Design Direction

**Instructions:**

Use the reference screenshot as the main source of visual inspiration rather than a strict layout template.

Preserve its most distinctive design characteristics, such as typography, visual rhythm, color relationships, component treatments, and overall aesthetic.

Feel free to modify section arrangements, proportions, spacing, and individual components to better suit my content.

Introduce original design decisions where they improve usability, composition, or brand consistency.

The result should clearly reflect the reference's visual language without looking like a direct reproduction.

Follow the visual identity of the reference design.

Use its typography, color palette, component styling, and overall aesthetic as the primary design direction.

Do not introduce an unrelated visual identity.

## Output

**Instructions:**

Generate a high-fidelity visual mockup of the requested interface.

Present the design as a polished, realistic website or product UI screenshot.

Prioritize visual composition, typography, spacing, colors, alignment, and component details.

Use the available image-generation or visual-design capabilities.

Do not replace the mockup with source code, a written design specification, or a textual description.

If direct visual generation is unavailable, clearly explain the limitation.

## Constraints

- Use the actual screenshot as the primary visual reference.
- Preserve essential messaging and factual accuracy.
- Prioritize visual quality and consistency.
- Do not invent unsupported product claims.
- Adapt structure and copy length for visual balance.

Reference imageAttach it to your message: assistants can’t open image links.

Build it in

Opening a tool copies the screenshot so you can paste it with your prompt.

Visit live page

Patterns

Filter
Search
Segmented Control
Status Indicator
Table

Tech Stack

React
Vercel

Added on

November 14, 2026

Deep Scan runs

The Deep Scan page lists every exploration run, named after what the agents investigated ("Note sharing and revocation flow", "Notes pinning and sidebar ordering", "Last note deletion and workspace persistence"), with a New Deep Scan button in the header.

Findings and coverage per run

Columns show the number of findings as a clickable pill (or "No findings reported"), coverage as a count of checks with a Completed or Partial status, the run duration and when it started.

Filtering

A search field, All Runs / With Findings toggles and a status select let teams jump straight to the scans that surfaced problems, with Clear Filters to reset.