More screenshots
Works with the AI tools you already use
AI UX Audit for Web & Mobile Apps
by Shandra
Audits web and mobile applications for usability, clarity, navigation, cognitive load, user-flow friction, CTA hierarchy, accessibility, responsive behavior, trust, error recovery, and conversion blockers. Produces evidence-based findings, severity and confidence ratings, prioritized recommendations, acceptance criteria, experiment ideas, QA checklists, and developer-ready implementation guidance.
Secure checkout via Stripe
See it in action
You say
Product name: FlowPilot
Product type: B2B SaaS project-management platform
Platforms: Responsive web application iOS companion app Android companion app
Primary users: Small-business owners Project managers Operations managers Freelancers Remote teams
User roles: Workspace Owner Administrator Project Manager Member Guest Collaborator
Primary user job: Create projects, assign work, understand progress, communicate with team members, and complete tasks on time.
Primary business goal: Increase workspace activation and improve paid-plan conversion.
Activation moment: A new workspace creates its first project, invites at least one collaborator, and completes the first assigned task.
Primary conversion: Start a paid subscription after the trial.
Critical user journeys: Landing page to signup Signup Workspace creation Team invitation First project creation Task assignment Dashboard review Search Notification management Plan comparison Subscription checkout Cancellation Account deletion
Current evidence: Desktop screenshots Mobile screenshots Figma prototype Analytics funnel Support-ticket summary Three usability-test recordings Current design-system documentation
Current problems: Users do not understand the difference between a workspace and a project. The dashboard shows too many cards. The primary CTA changes between screens. New users are sent to an empty dashboard. Team invitation appears before the product value is clear. Mobile navigation contains seven destinations. Search filters reset after viewing a result. Forms validate only after submission. Some error messages say “Something went wrong.” The trial page does not clearly explain renewal. Cancellation is hidden inside Billing. The mobile keyboard covers the Save button. Several controls use icons without labels. Low-contrast gray text is used for important status information. The settings area mixes personal and workspace preferences. Analytics track page views but not successful task completion.
Target markets: United States United Kingdom European Union Canada Australia
Accessibility target: WCAG 2.2 AA-oriented implementation review Keyboard support Screen-reader support 200% zoom Large text Reduced motion Mobile-platform accessibility review
Technology stack: Next.js React TypeScript Tailwind CSS shadcn/ui React Native Existing design system
Need output: Executive UX audit Critical findings First-impression analysis Information architecture Navigation audit Onboarding audit Dashboard audit Task-creation flow Team-invitation flow Search and filter audit Form audit Error and recovery audit Subscription audit Settings audit Mobile-usability audit Accessibility audit Conversion blockers Severity and confidence scoring Quick wins Strategic recommendations Analytics event plan Experiment roadmap Developer handoff Acceptance criteria Regression tests QA checklist Risk assessment
Need variations: Executive Stakeholder Report Detailed Designer Audit Developer Implementation Backlog
Special constraints: Do not invent analytics. Do not claim conversion impact is proven. Do not invent subscription terms. Do not invent user-research findings. Do not claim accessibility conformance. Do not recommend hiding cancellation. Do not use manipulative urgency. Do not prioritize conversion over user safety or accessibility.
Your agent does
=== AI UX AUDIT FOR WEB & MOBILE APPS ===
Product: FlowPilot
Product Type: B2B SaaS project-management platform
Platforms: Responsive web, iOS, and Android
Primary User Job: Create projects, coordinate work, understand progress, and complete tasks with a team.
Activation Moment: Create the first project, invite a collaborator, and complete the first assigned task.
=== EXECUTIVE SUMMARY ===
FlowPilot provides the essential features required for team-based project management, but the current experience creates substantial friction before users reach the activation moment.
The most important issues are:
- Workspace and project terminology are not explained clearly.
- New users are routed to an empty dashboard without a guided first task.
- The dashboard presents too many competing modules.
- Team invitation is requested before users understand product value.
- CTA labels and placement are inconsistent.
- Mobile navigation exposes too many top-level destinations.
- Search state is lost after viewing a result.
- Form errors appear too late.
- Generic error messages do not support recovery.
- Trial renewal terms are incomplete.
- Cancellation is difficult to discover.
- Personal and workspace settings are mixed.
- Mobile keyboard behavior obstructs important actions.
- Important status text has insufficient visual emphasis.
- Analytics do not measure meaningful task completion.
The recommended priority is to simplify the activation journey before redesigning secondary areas.
Recommended activation flow:
Signup → Workspace Setup → Create First Project → Add First Task → Invite Collaborator in Context → Complete First Task → Activation Confirmation → Dashboard
=== AUDIT SCORING ===
Clarity: Weak
Navigation: Weak
Activation Flow: Critical
Dashboard Hierarchy: Weak
Forms: Adequate
Error Recovery: Weak
Mobile Usability: Weak
Accessibility Readiness: Weak
Subscription Transparency: Critical
Settings Architecture: Weak
Analytics Readiness: Weak
Overall: Weak
The overall score is a heuristic assessment and does not represent validated user performance.
=== CRITICAL FINDING 1 ===
Finding: New users land on an empty dashboard without a guided activation path.
Evidence: The supplied first-use dashboard contains empty widgets, navigation links, and a general Create button, but no explanation of the recommended first task.
User Impact: New users must infer the product model and decide what to create without guidance.
Business Impact: This likely delays activation and may increase first-session abandonment.
Confidence: High
Severity: Critical
Recommendation: Replace the generic first-use dashboard with a guided activation state.
Recommended sequence:
- Create a project.
- Add the first task.
- Invite a collaborator.
- Complete or update the task.
- Show activation completion.
Acceptance Criteria:
Given a newly created workspace has no projects, when the owner opens the dashboard, then the interface presents one primary action to create the first project and explains how that action begins the setup journey.
=== CRITICAL FINDING 2 ===
Finding: The trial page does not communicate renewal clearly.
Evidence: The supplied trial interface displays “Start Free Trial” and plan benefits, but the future billing cadence, renewal behavior, and verified post-trial price are not visible in the provided screen.
User Impact: Users cannot fully understand the commitment.
Trust Impact: The incomplete information may reduce confidence and create billing disputes.
Severity: Critical
Confidence: High
Recommendation: Display verified price, currency, billing cadence, trial duration, renewal behavior, included features, limitations, and cancellation access before the final action.
Requires Verification: Exact commercial terms.
=== HIGH FINDING 1 ===
Finding: Workspace and project terminology compete without explanation.
Evidence: Navigation, onboarding, and dashboard screens use both terms, but no supplied interface explains their relationship.
User Impact: Users may create content at the wrong level or misunderstand where members, billing, and settings apply.
Severity: High
Confidence: High
Recommendation: Introduce a concise product model during workspace setup and reinforce scope labels in relevant interfaces.
Example:
Workspace: The team environment containing members, billing, and projects.
Project: A collection of tasks and activity inside a workspace.
Use final wording only after product review.
=== HIGH FINDING 2 ===
Finding: The dashboard has no dominant task.
Evidence: The supplied desktop dashboard displays eight cards with similar visual weight.
User Impact: Users must scan multiple areas before understanding what requires attention.
Cognitive-Load Impact: High.
Recommendation: Prioritize dashboard content in this order:
- Current urgent state
- Resume active work
- Primary task
- Assigned work
- Project progress
- Secondary insights
Reduce equal-weight cards and demote low-frequency content.
=== HIGH FINDING 3 ===
Finding: Team invitation occurs before sufficient product value.
Evidence: The onboarding flow requests invitations immediately after workspace creation and before project creation.
User Impact: Users must involve colleagues before understanding the product.
Recommendation: Make invitation skippable and move the strongest invitation prompt into project setup or task assignment.
Contextual invitation:
Assign this task to a teammate.
Secondary action:
Continue on my own.
=== NAVIGATION AUDIT ===
Desktop Navigation:
Current: Dashboard My Work Projects Calendar Messages Reports Templates Automations Settings
Observed Problem: Several destinations compete at the same level before the user understands the product.
Recommendation: Group secondary destinations and preserve the most frequent destinations.
Possible structure:
Home My Work Projects Inbox More
The final model requires usage data and role review.
Mobile Navigation:
Current: Seven bottom destinations
Severity: High
Recommendation: Use a smaller set of stable top-level destinations and move lower-frequency areas into a secondary menu.
Possible mobile tabs:
Home My Work Projects Inbox Account
Do not finalize navigation without analytics and user validation.
=== CTA AUDIT ===
Observed CTA Labels: Create Add Continue Next Done Submit
Problem: The same action type uses different labels, and several labels do not describe the result.
Recommendations:
Create: Replace with object-specific actions such as Create Project or Add Task.
Continue: Use when the next step is predictable.
Submit: Replace with the actual result, such as Send Invitation or Save Changes.
Done: Use only when the task is complete and no further action is implied.
Primary CTA Placement: Keep the primary action in a stable location within repeated flows.
Mobile: Ensure the action remains visible when the keyboard is open without covering form content.
=== SEARCH AND FILTER AUDIT ===
Finding: Search state is lost after viewing a project.
Observed: Returning from project detail clears the query, filters, and result position.
Impact: Users must repeat work.
Severity: High
Recommendation: Preserve:
Search query Applied filters Sort order Result position Selected tab
Acceptance Criteria:
Given a user opens a project from filtered search results, when the user returns, then the query, filters, sorting, and result position are restored.
=== FORM AUDIT ===
Observed Problems:
- Validation occurs after final submission.
- Required fields are not consistently identified.
- Some placeholders are used instead of persistent labels.
- Error summaries are missing.
- Mobile Save actions can be obstructed by the keyboard.
Recommendations:
- Use persistent labels.
- Identify optional fields rather than marking every required field.
- Validate after sufficient interaction.
- Preserve valid values after failure.
- Associate errors programmatically with fields.
- Move focus to the error summary or first invalid field according to the approved accessibility pattern.
- Keep mobile actions keyboard-safe.
=== ERROR AND RECOVERY AUDIT ===
Current Message: Something went wrong.
Problem: The message does not identify the failed action or recovery path.
Recommended Formula:
What happened + what remains safe + what the user can do
Example:
We could not save this task. Your changes are still visible on this screen. Check your connection and try again.
Use the preservation statement only when technically verified.
Required Error Categories:
- Validation
- Permission
- Network
- Server
- Conflict
- Session expired
- Upload
- Subscription
- Unknown
=== MOBILE-USABILITY AUDIT ===
High-Priority Problems:
- Seven bottom-navigation items
- Keyboard covers Save
- Dense project tables
- Small icon-only controls
- Status conveyed primarily through color
- Filters occupy most of the viewport
- Dialogs do not adapt to short screens
Recommendations:
- Reduce top-level mobile destinations.
- Convert dense tables into task-oriented cards.
- Use labeled controls for important actions.
- Keep primary actions above the keyboard.
- Use non-color status indicators.
- Move complex filters into an accessible full-height sheet.
- Test large text and landscape orientation.
- Avoid page-level horizontal scrolling.
=== ACCESSIBILITY AUDIT ===
Observed or Likely Risks:
- Icon-only controls lack visible labels.
- Important status text has low contrast.
- Form errors are not associated with fields.
- Modal focus behavior is not documented.
- Mobile touch targets appear inconsistent.
- Color is used as the primary status indicator.
- The dashboard hierarchy may be difficult to navigate with screen-reader headings.
Recommendations:
- Provide accessible names.
- Add text labels for high-value actions.
- Verify contrast.
- Use semantic headings.
- Associate errors with inputs.
- Define focus entry and return for dialogs.
- Increase touch targets.
- Add text or icon indicators in addition to color.
- Test at 200% zoom and large text.
- Test with keyboard and screen readers.
Do not claim conformance until implementation testing is complete.
=== SETTINGS AUDIT ===
Finding: Personal and workspace settings are mixed.
Examples: Personal language, workspace billing, personal security, and team notifications appear in one category list.
Recommendation:
Personal:
- Profile
- Security
- Personal Notifications
- Language
- Accessibility
Workspace:
- Members
- Roles
- Billing
- Integrations
- Workspace Notifications
- Data
Always display the active settings scope.
=== CONVERSION BLOCKERS ===
Critical:
- Unclear activation path
- Incomplete trial-renewal information
High:
- Premature invitation
- Dashboard overload
- Search-state loss
- Mobile navigation overload
- Hidden cancellation
Medium:
- Generic CTA labels
- Late form validation
- Weak error recovery
- Inconsistent status presentation
These impacts are hypotheses unless supported by behavioral data.
=== QUICK WINS ===
- Replace generic first-use dashboard with a guided empty state.
- Clarify workspace and project terminology.
- Make team invitation skippable.
- Use object-specific CTA labels.
- Preserve search and filter state.
- Add field-level validation.
- Replace generic error copy.
- Keep mobile Save actions keyboard-safe.
- Separate personal and workspace settings.
- Make subscription cancellation discoverable.
- Add text to critical icon-only controls.
- Improve status contrast and non-color communication.
=== STRATEGIC RECOMMENDATIONS ===
- Redesign the activation journey around the first completed team task.
- Simplify dashboard hierarchy by user role and current state.
- Rebuild navigation around high-frequency jobs.
- Establish a product-wide CTA and feedback system.
- Define reusable error, loading, empty, and success states.
- Build an accessibility baseline into the design system.
- Instrument task completion rather than page views alone.
- Separate personal, workspace, and organization settings.
- Create transparent subscription and cancellation flows.
- Validate the redesign through usability testing and controlled experiments.
=== ANALYTICS EVENT PLAN ===
Event: workspace_created
Purpose: Measure the beginning of setup.
Event: first_project_created
Properties:
- entry_source
- template_used
- platform
Event: first_task_created
Properties:
- project_type
- platform
Event: first_collaborator_invited
Properties:
- invitation_context
- skipped_initial_invitation
Event: first_task_completed
Properties:
- time_bucket
- collaboration_state
Event: search_result_opened
Properties:
- query_length_bucket
- filter_count
- result_position_bucket
Event: form_validation_failed
Properties:
- form_id
- error_category
- field_category
Event: subscription_viewed
Properties:
- entry_source
- current_plan
- user_role
Privacy: Do not record private task content, message content, passwords, payment details, uploaded files, or sensitive profile values.
=== EXPERIMENT ROADMAP ===
Experiment 1: Guided First-Use Dashboard
Hypothesis: A task-based first-use state will increase completion of the first project.
Primary Metric: First project created
Guardrail: Setup abandonment
Experiment 2: Contextual Invitation
Hypothesis: Requesting invitations after project creation will increase qualified invitations without delaying activation.
Primary Metric: Collaborator invited within the activation journey
Guardrail: First task completion
Experiment 3: Reduced Dashboard Modules
Hypothesis: Prioritizing resume and assigned work will improve task initiation.
Primary Metric: Task opened from dashboard
Guardrail: Secondary-feature discovery
Experiment 4: Transparent Trial Summary
Hypothesis: Displaying complete verified billing terms will improve qualified trial starts and reduce billing-related support contacts.
Primary Metric: Trial start
Guardrails: Cancellation Billing support contact
=== DEVELOPER HANDOFF ===
Affected Areas:
- onboarding
- dashboard
- navigation
- project creation
- task creation
- invitation flow
- search
- forms
- error system
- subscription
- settings
- mobile layout
- accessibility
- analytics
Required Shared Components:
- GuidedEmptyState
- PrimaryAction
- SecondaryAction
- ErrorSummary
- FieldError
- SystemStatusBanner
- MobileActionBar
- AccessibleDialog
- FilterSheet
- ScopeIndicator
- SubscriptionSummary
- ConfirmationState
Required State Coverage:
- default
- loading
- empty
- success
- validation error
- permission error
- network error
- conflict
- session expired
- offline
- partial success
=== ACCEPTANCE CRITERIA ===
Activation:
Given a new workspace has no projects, when the owner enters the product, then one clear primary action guides the owner toward creating the first project.
Invitation:
Given a user has not invited a collaborator, when the first project is created, then the product offers a contextual invitation while preserving a clear option to continue alone.
Search:
Given a user opens a result from filtered search, when the user returns, then the query, filters, sort order, and result position remain available.
Form Validation:
Given a user enters invalid information, when validation occurs, then the affected field is identified, the error is programmatically associated, and valid values remain intact.
Subscription:
Given a user reviews the trial, when the final action is presented, then verified price, currency, trial duration, billing cadence, renewal, limitations, and cancellation access are visible.
Mobile Keyboard:
Given the virtual keyboard is open, when a user completes a form, then the primary action remains reachable and does not cover required content.
Accessibility:
Given the interface is used at 200% zoom, when content reflows, then the user can navigate and complete primary flows without page-level horizontal scrolling.
=== REGRESSION TEST MATRIX ===
Platforms:
- Desktop Web
- Tablet Web
- Mobile Web
- iOS
- Android
Roles:
- Owner
- Administrator
- Project Manager
- Member
- Guest
States:
- New Workspace
- Active Workspace
- Empty Project
- Populated Project
- Trial
- Paid
- Payment Failed
- Restricted
Flows:
- Signup
- Workspace Setup
- Project Creation
- Task Creation
- Invitation
- Search
- Subscription
- Cancellation
- Settings
- Deletion
Accessibility:
- Keyboard
- Screen Reader
- 200% Zoom
- Large Text
- Reduced Motion
- Non-Color Status
- Landscape Mobile
=== QA CHECKLIST ===
Clarity:
- Verify page purpose.
- Verify primary action.
- Verify terminology.
- Verify labels.
- Verify instructions.
Navigation:
- Verify active state.
- Verify Back behavior.
- Verify mobile navigation.
- Verify workspace switching.
- Verify deep-link return.
Flows:
- Verify entry.
- Verify progress.
- Verify skip.
- Verify resume.
- Verify success.
- Verify recovery.
Forms:
- Verify labels.
- Verify autofill.
- Verify validation.
- Verify error relationships.
- Verify keyboard behavior.
- Verify data preservation.
States:
- Verify loading.
- Verify empty.
- Verify error.
- Verify offline.
- Verify success.
- Verify partial success.
Accessibility:
- Verify semantics.
- Verify focus.
- Verify keyboard operation.
- Verify screen-reader order.
- Verify contrast.
- Verify touch targets.
- Verify zoom and reflow.
- Verify reduced motion.
Mobile:
- Verify small screens.
- Verify safe areas.
- Verify keyboards.
- Verify bottom sheets.
- Verify orientation.
- Verify large text.
- Verify no obstructed actions.
=== RISKS ===
Risk: Recommendations are treated as proven conversion facts.
Mitigation: Label conversion effects as hypotheses until validated.
Risk: Navigation is simplified without role analysis.
Mitigation: Validate destination frequency and role requirements.
Risk: Accessibility improvements are postponed.
Mitigation: Include accessibility requirements in component acceptance criteria.
Risk: Activation optimization creates unwanted pressure.
Mitigation: Preserve skip, recovery, transparency, and user control.
Risk: Analytics capture confidential product content.
Mitigation: Track event categories and outcomes rather than private user data.
Risk: Mobile fixes are validated only in desktop browser emulation.
Mitigation: Test on real devices, operating systems, keyboards, and assistive technologies.
=== KNOWN LIMITATIONS ===
- Static screens cannot reveal every interaction problem.
- User impact requires research or behavioral evidence.
- Conversion impact remains a hypothesis without reliable data.
- Accessibility requires implementation testing.
- Technical feasibility requires engineering review.
- Localization requires professional in-context review.
- Analytics quality depends on correct instrumentation.
- Experiments require sufficient traffic and statistical discipline.
- The audit does not guarantee conversion, retention, revenue, accessibility conformance, security, or product success.
=== FINAL INTEGRITY NOTE ===
Audit findings must distinguish direct evidence from inference.
Do not claim that a recommendation will increase conversion, activation, retention, or revenue unless reliable evidence supports that conclusion.
Do not claim accessibility conformance, security, payment success, data preservation, or production readiness without implementation verification and testing.
The audit is designed to identify usability risks, clarify priorities, and improve implementation readiness. It does not replace moderated user research, accessibility auditing, engineering validation, security review, legal review, analytics validation, or production QA.
What you get
About this skill
AI UX Audit for Web & Mobile Apps helps startup founders, SaaS teams, product designers, UI/UX agencies, developers, product managers, growth teams, and no-code creators identify the interface problems that confuse users, interrupt important journeys, reduce task completion, and weaken conversion.
The skill supports responsive websites, web applications, mobile applications, SaaS platforms, marketplaces, e-commerce stores, booking systems, subscription products, dashboards, onboarding experiences, internal tools, productivity applications, fintech interfaces, education platforms, AI products, membership systems, creator tools, and multi-step transactional flows.
It can audit an interface from screenshots, screen recordings, wireframes, prototypes, Figma exports, live-product descriptions, analytics summaries, support tickets, usability notes, product requirements, source-code excerpts, or structured screen inventories.
The audit begins by defining:
Product type Target audience Primary user roles Primary user job Primary business objective Activation moment Primary conversion Critical user journeys Target platforms Target devices Accessibility requirements Known business rules Known technical constraints Available evidence
The agent evaluates the experience from the user's perspective rather than judging screens only by visual style.
It asks:
Can users understand what the product does?
Can users identify the correct next action?
Can users complete the primary task without unnecessary effort?
Does the interface explain system status?
Can users recover from mistakes?
Are important actions discoverable?
Are risks and consequences understandable?
Does the experience work on mobile?
Can users with disabilities operate it?
Does the interface preserve trust?
Are conversion opportunities supported without manipulation?
The skill creates evidence-based UX findings rather than generic design opinions.
Every important finding can include:
Affected screen Affected flow Observed interface evidence User problem Likely user behavior Business impact Accessibility impact Trust impact Severity Priority Effort Confidence Recommendation Validation method Acceptance criteria
The audit clearly distinguishes:
Observed: Directly visible or supported by screenshots, recordings, code, analytics, user research, product requirements, or verified behavior.
Inferred: A likely usability or conversion problem suggested by available evidence but not directly confirmed through user testing or analytics.
Recommended: A proposed UX, UI, content, accessibility, navigation, flow, or implementation improvement.
Requires Verification: Any business rule, payment behavior, subscription condition, security state, legal requirement, technical limitation, data behavior, user assumption, or causal claim that must be confirmed.
The skill can analyze first impressions and product comprehension.
First-impression analysis can evaluate:
Product category recognition Primary value proposition Audience relevance Visual focus Page purpose Primary action Trust signals Content density Competing messages Time to comprehension Mismatch between marketing promise and product experience
The agent identifies whether users can understand the interface quickly without reading excessive text or exploring several unrelated screens.
Clarity analysis covers:
Headlines Labels Instructions Navigation names Form fields Buttons Status messages Empty states Error messages Confirmation messages Product terminology Technical language Ambiguous icons Hidden assumptions Unclear abbreviations Inconsistent naming
The skill detects cases where the interface uses internal product terminology instead of language users understand.
Information-architecture analysis can evaluate:
Content grouping Category structure Settings organization Page hierarchy Navigation depth Feature discoverability Role-based destinations Workspace scope Account scope Repeated destinations Dead ends Orphaned screens Overlapping categories Poorly named sections Excessive nesting Missing cross-links
Navigation analysis covers:
Global navigation Local navigation Mobile navigation Bottom tabs Sidebars Drawers Breadcrumbs Search-led navigation Back behavior Modal behavior Deep links Active states Navigation persistence Role-specific navigation Workspace switching Return paths External handoffs
The agent identifies whether users can answer:
Where am I?
How did I get here?
What can I do here?
How do I go back?
What will happen when I continue?
User-flow analysis can cover:
Onboarding Signup Login Account recovery Account setup Permissions Dashboard entry Search Filtering Content creation File upload Checkout Booking Subscription Payment Profile Settings Notifications Data export Cancellation Account deletion Support Error recovery
Each flow can be reviewed for:
Entry points Prerequisites Step count Required effort Progress visibility Decision complexity Interruptions Alternate paths Error states Save and resume Exit behavior Success confirmation Post-success guidance
The skill identifies unnecessary steps, premature commitment, repeated data entry, missing progress indicators, weak transitions, and dead-end success states.
CTA analysis evaluates:
Primary CTA visibility Secondary CTA hierarchy Competing actions CTA labels Action specificity Destructive actions Disabled actions Sticky actions Mobile action placement CTA repetition CTA timing Context State changes Post-click feedback
The agent distinguishes destination labels from action labels and detects vague buttons such as:
Continue Submit Next Proceed Done Learn More
when more specific language would improve confidence.
Cognitive-load analysis evaluates:
Number of visible decisions Information density Competing priorities Visual noise Form length Complex terminology Memory requirements Repeated explanations Unnecessary options Decision fatigue Context switching Hidden dependencies Premature configuration Feature overload
Recommendations may include progressive disclosure, grouping, sensible defaults, contextual education, task segmentation, reduced choice, simplified content, or clearer prioritization.
Visual-hierarchy analysis can review:
Headline prominence CTA prominence Content sequence Typography Spacing Alignment Contrast Grouping Card hierarchy Color usage Illustration priority Image relevance Table density Chart emphasis Mobile stacking Empty-space balance
The skill does not equate visual minimalism with good UX. It evaluates whether the hierarchy supports the user's current task.
Content and microcopy analysis covers:
Headlines Descriptions Labels Helper text Placeholder text Button copy Validation messages Error copy Warnings Confirmation messages Empty states Loading messages Permission explanations Subscription copy Cancellation copy Deletion copy Support copy
The agent prioritizes plain language, accurate expectations, concise instructions, and actionable recovery.
Form analysis can evaluate:
Required and optional fields Field grouping Input types Autofill Labels Placeholder misuse Validation timing Error placement Error recovery Field dependencies Address formats Phone formats Password requirements Date entry File upload Long forms Multi-step forms Save and resume Virtual keyboards Mobile focus order Submission state
The skill detects form patterns that create abandonment, such as unnecessary fields, unclear requirements, validation only after submission, erased values, inaccessible errors, and keyboard-obstructed actions.
Error-state analysis covers:
Validation errors Authentication failures Permission failures Network errors Server errors Payment problems Upload failures Synchronization conflicts Unavailable content Expired sessions Booking conflicts Inventory changes Subscription states Unknown failures
An effective error should communicate:
What happened What was preserved What the user can do Whether retrying is safe Where to get help Whether the problem affects other work
The skill does not claim data was preserved unless that behavior is verified.
Recovery analysis can identify whether users can:
Retry safely Edit incorrect information Resume interrupted work Restore a draft Change a payment method Reconnect an integration Return to a previous step Contact support Cancel without losing unrelated progress Understand uncertain system states
Loading and system-status analysis covers:
Page loading Component loading Form submission File upload AI generation Payment processing Background synchronization Streaming Long-running operations Empty loading regions Skeleton states Progress indicators Cancellation Timeouts
The agent avoids fake percentages, unsupported completion times, and loading states that remove useful context.
Empty-state analysis can review:
First use No search results Filtered empty No data No activity Completed tasks Permission-restricted content Unavailable content Archived content Expired content
A useful empty state should explain what belongs in the area, why it is empty, and which meaningful action is available.
Trust analysis covers:
Pricing transparency Subscription terms Trial conditions Renewal Cancellation Refund context Security language Privacy language Data use User control Seller identity Customer support Reviews Guarantees Permissions Destructive actions AI limitations System status
The agent rejects:
Fake urgency Artificial scarcity Confirm-shaming Hidden cancellation Hidden deletion Preselected marketing consent Misleading discounts Unsupported guarantees False security claims Dark patterns Deceptive button hierarchy Disguised advertising Unexpected recurring charges
Accessibility analysis can cover:
Semantic structure Heading hierarchy Landmarks Accessible names Persistent labels Alternative text Keyboard navigation Visible focus Focus order Focus management Screen-reader order Error relationships Touch targets Contrast Text resizing Zoom Reflow Reduced motion Non-color status Accessible dialogs Accessible drawers Accessible bottom sheets Accessible tables Accessible charts Form instructions Timeouts Orientation External keyboards Gesture alternatives Large text Mobile assistive technologies
The skill may recommend WCAG AA-oriented implementation review when no other target is supplied, but it does not claim conformance without implementation testing.
Mobile-usability analysis covers:
Small-screen hierarchy Thumb reach Touch-target size Bottom navigation Sticky controls Safe areas Mobile keyboards Viewport height Horizontal scrolling Responsive tables Dense filters Bottom sheets Gestures Orientation Large text Mobile search Mobile forms Back behavior Interrupted sessions Native-platform conventions
The agent identifies interfaces that were designed for desktop and merely compressed for mobile.
Responsive analysis can define behavior for:
Small mobile Standard mobile Large mobile Tablet Laptop Desktop Wide desktop Short viewport Landscape mobile 200% zoom Large-text mode
Conversion-blocker analysis can cover:
Unclear value Weak CTA hierarchy Forced signup Premature permission requests Hidden costs Long checkout Poor trust Form friction Navigation confusion Missing proof Unclear plan differences Weak onboarding Uncertain payment Incomplete confirmation Inaccessible controls Mobile failures Poor recovery Subscription ambiguity Cancellation distrust
The agent distinguishes usability issues from conversion hypotheses. It does not present an inferred conversion effect as proven fact.
Onboarding audits can evaluate:
Value introduction Account-creation timing Setup length Progress Skip behavior Permissions Personalization First task Activation Completion Resume behavior Contextual education
Authentication audits can evaluate:
Signup Login Social login Magic links One-time codes Password reset Verification Account conflicts Duplicate accounts Recovery Session expiration Deep-link return
Dashboard audits can evaluate:
Current status Primary task Urgent information Resume actions Recent activity Recommendations Data density Widget priority Empty states Role differences Customization Mobile behavior
Search and filtering audits can evaluate:
Search entry Suggestions Recent searches Query preservation Filters Sorting Result relevance No results Filter visibility Applied-filter states Reset behavior Mobile filtering Result-position preservation
Checkout and payment audits can evaluate:
Cart review Guest checkout Shipping Billing Discounts Payment methods Pricing Taxes Fees Trust Validation Declines Uncertain payment Duplicate prevention Confirmation Abandoned-checkout recovery
Subscription audits can evaluate:
Plan comparison Pricing Billing cadence Trial Renewal Usage limits Upgrade Downgrade Cancellation Expiration Reactivation Payment failure
Profile and settings audits can evaluate:
Personal information Public profile Private profile Security Sessions Billing Notifications Privacy Connected accounts Preferences Data export Deactivation Deletion
The skill can create scoring systems based on:
Severity User impact Business impact Frequency Reach Confidence Effort Accessibility risk Trust risk Implementation dependency
Severity levels can include:
Critical: Blocks completion, creates severe risk, causes data or payment uncertainty, or makes an essential flow unusable.
High: Creates substantial friction, confusion, abandonment risk, accessibility failure, or trust damage.
Medium: Reduces efficiency, clarity, confidence, or consistency without fully blocking the task.
Low: Creates minor friction, polish issues, or limited inconsistency.
Confidence levels can include:
High: Directly visible, repeatedly evidenced, or supported by reliable data.
Medium: Strongly suggested but requires validation.
Low: Plausible hypothesis with limited evidence.
Prioritization can combine:
Impact Reach Confidence Effort Risk Dependency Strategic importance
The skill can generate:
Executive UX summaries Critical-issue reports Screen-by-screen audits Flow-by-flow audits Heuristic evaluations Accessibility reviews Mobile-usability reviews Conversion-blocker reports Trust-risk reports Cognitive-load reports CTA audits Form audits Checkout audits Onboarding audits Settings audits Before-and-after recommendations Prioritized roadmaps Quick-win lists Strategic redesign recommendations Analytics event plans Experiment roadmaps Developer handoffs Acceptance criteria Regression tests QA checklists Structured JSON specifications
Analytics recommendations can define events for:
Flow entry Step completion CTA activation Form error Validation failure Search Filter Empty state Permission response Checkout progress Payment state Subscription state Cancellation Error recovery Support contact Task completion
Analytics must avoid passwords, private messages, payment credentials, health information, sensitive profile details, raw uploaded content, and confidential account data.
Experiment recommendations can test:
CTA copy CTA position Step count Progress indicators Navigation structure Form length Guest checkout Permission timing Trust placement Pricing presentation Plan comparison Error messaging Mobile layout Search filters Onboarding order Dashboard priority
Each experiment should define:
Observation Hypothesis Control Variation Primary metric Guardrail Audience Duration logic Sample requirement Risk Decision rule
The skill produces developer-ready recommendations rather than abstract statements such as “make it cleaner” or “improve UX.”
Implementation guidance can include:
Affected route Affected component Current state Target state Responsive behavior Interaction behavior Loading Error Empty state Accessibility Analytics Acceptance criteria Dependencies Open questions
It prioritizes evidence, user-task completion, clarity, accessibility, mobile usability, trust, recovery, and implementation feasibility.
The core commercial promise is: identify the UX problems that users experience but product teams frequently overlook, then convert those findings into prioritized, testable, implementation-ready improvements.
How to install
Drop the file into your AI Agent. Works with Claude, Cursor, ChatGPT, and 20+ more.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill passes our trust checks below.
Security scanned
Passed our 8-point scan before listing
Fresh listing
Recently published to Agensi
30-day refund
Not a fit? Get your money back
Trust & safety
Security scanned
Verified clean today
- 30-day refund guarantee
- One-time purchase, yours forever
- Secure checkout via Stripe
Creator
Shandra is a top-ranked AI prompt creator and premium agent skill builder with an established track record in the AI marketplace. She is recognized as a #1 Top Seller on PromptBase, where she has built a trusted catalog of specialized AI prompts and agent skills for creators, entrepreneurs, educators, marketers, digital product sellers, and business professionals. With over 3,500 AI products published, more than 3,400 sales, and 1,000+ five-star reviews, Shandra has become known for creating practical, polished, and commercially useful AI resources that help users save time, organize complex ideas, generate high-quality content, build digital products, and transform creative concepts into actionable workflows. Her Agensi store focuses on premium, ready-to-use agent skills designed for real-world productivity. Each skill is developed with clear instructions, structured workflows, professional formatting, practical use cases, setup guidance, examples, edge-case handling, and a strong emphasis on usability. Her work combines creative strategy, prompt engineering, documentation design, business thinking, and practical automation into reliable tools that users can apply immediately. Shandra’s mission is to create AI skills that feel professional, useful, and complete from the first use — not generic templates, but carefully built workflow systems that help users think better, work faster, and produce stronger results.
Frequently Asked Questions
Popular in Marketing & SEO
keyword-research
Transform URLs or product lists into SEO keyword research packs with Google Ads data and intent-based clustering.
local-seo-audit
Automated local SEO auditing and keyword generation for agencies and small businesses.
click-to-convert-audit
High-integrity landing page audits that identify ad spend hazards and conversion blockers for Google Ads traffic.
content architect pro
Analyze, audit, and optimize website content with AI-powered content strategy, UX writing reviews, SEO insights, and conversion-focused copywriting