Web UI Microcopy & Interaction States Agent

Web UI Microcopy & Interaction States Agent helps UX writers, UI designers, SaaS teams, product managers, frontend developers, design-system teams, agencies, and product organizations create interface language that makes digital products easier to understand, trust, and operate.

The skill transforms visually polished but unclear interfaces into coherent product experiences by defining the words, states, transitions, explanations, confirmations, warnings, and recovery guidance users encounter before, during, and after interaction.

It can create microcopy and interaction-state systems for SaaS products, e-commerce stores, AI applications, admin panels, onboarding flows, signup experiences, checkout systems, booking platforms, marketplaces, subscription products, internal tools, dashboards, account settings, billing portals, file-management interfaces, collaboration products, mobile web applications, enterprise software, and customer-support workflows.

The agent begins by identifying the user goal, business goal, component context, system behavior, operational risk, product terminology, brand voice, and available recovery paths. It writes only after distinguishing what the system knows, what remains uncertain, what changed, what was preserved, whether an operation is reversible, and what the user can do next.

It creates terminology systems that define preferred product terms, restricted words, synonyms, entity names, action names, status labels, capitalization rules, punctuation conventions, and usage examples. This prevents different screens from referring to the same concept with inconsistent language.

The skill defines brand voice and context-sensitive tone. Routine actions may use neutral and concise language, onboarding may be supportive, security and billing states should remain calm and precise, destructive actions should be direct, and success states should be positive without becoming vague or excessively celebratory.

It creates specific button labels based on the action and outcome. Generic labels such as Submit, Continue, Done, Yes, and OK are replaced with clearer alternatives such as Request a Quote, Create Workspace, Continue to Payment, Confirm Booking, Archive Project, or Delete Account.

For each action, the agent can specify the default label, loading label, success response, error response, disabled state, permission state, mobile treatment, and accessible name.

The skill creates link labels that explain the destination or outcome, accessible labels for icon-only controls, navigation labels, menu items, tabs, filters, sorting controls, pagination actions, and table actions.

It designs form microcopy including persistent labels, required and optional conventions, helper text, examples, privacy explanations, format guidance, character-limit messages, password requirements, file-upload instructions, consent language structure, and specific submission actions.

Validation messages explain what is wrong and how to correct it without blaming the user. Validation timing is selected according to the context, including during input, after leaving a field, at step completion, or on submission.

The agent creates complete error architecture for field errors, form errors, network failures, server errors, timeouts, permission problems, authentication failures, payment failures, unavailable booking slots, upload failures, import failures, export failures, duplicate actions, plan limits, rate limits, moderation blocks, partial failures, offline states, and unknown failures.

Error messages can explain what happened, what information or work was preserved, whether the action was confirmed, whether retrying is safe, what the primary recovery action is, and where additional support is available.

The skill distinguishes full failure from partial success. For bulk actions, imports, uploads, exports, or multi-item operations, it can communicate completed records, failed records, excluded records, reasons, retry paths, and review actions.

It creates first-use empty states, filtered empty states, no-search-result states, completed states, permission-restricted states, archived states, expired states, and unavailable states.

First-use empty states can explain what belongs in the section, why it matters, how to begin, and where to find examples or templates. Filtered empty states can help users clear filters or modify their search without implying that the product contains no data.

Loading and progress systems can cover initial page loading, component loading, button actions, uploads, file processing, report generation, payment processing, background tasks, pagination, streaming, AI generation, synchronization, and imports.

The agent avoids fake percentages and unsupported timing claims. Determinate progress is used only when progress can actually be measured. Indeterminate states use accurate language such as Saving, Processing the File, Preparing the Report, or Generating the Result.

Success messages explain what completed, what changed, what happens next, whether further action is required, and whether Undo is available.

The skill creates toast systems, inline confirmations, banners, page-level confirmations, confirmation screens, notification messages, and background-completion states. Critical information is not placed only inside temporary toasts.

It designs warnings that explain risk, consequence, and the safest next step. It separates warnings from errors and neutral information so users can accurately interpret severity.

Confirmation dialogs are designed according to risk. Low-risk actions may use concise confirmation or Undo. Medium-risk actions explain consequences. High-risk actions may require scope details, affected users, reversibility, reason capture, typed confirmation, secondary approval, or delayed execution.

Destructive-action copy explains what will be deleted, removed, suspended, cancelled, disconnected, or changed; which users or resources are affected; when the action takes effect; what remains; and whether recovery is possible.

The skill creates tooltips for concise secondary explanations, unfamiliar icons, keyboard shortcuts, and brief definitions. Essential instructions, legal terms, error recovery, and information required to complete a task remain visible rather than being hidden inside tooltips.

Onboarding microcopy can include welcome messages, first-use hints, contextual guidance, checklists, coach marks, progressive education, setup steps, completion states, skip actions, resume states, and first-value guidance.

Onboarding hints are designed to teach meaningful tasks rather than merely naming interface features. They appear at relevant moments, remain dismissible, avoid blocking the interface, and do not repeatedly reappear after completion.

Notification architecture can cover success, warnings, errors, action-required states, system status, collaboration events, billing problems, security alerts, account issues, and background completion.

Banner systems can communicate platform incidents, account restrictions, payment failures, trial status, maintenance, impersonation, permission scope, and important product changes.

The agent creates complete upload-state copy for idle, dragging, validating, uploading, processing, ready, unsupported, oversized, duplicate, failed, cancelled, and removed states.

It can also create import and export state systems covering validation, mapping, preview, processing, completion, partial completion, failure, expiration, rollback, permission restrictions, and downloadable results.

For search, filters, sorting, and tables, the skill creates placeholders, search guidance, no-result messages, searchable-field explanations, applied-filter labels, clear actions, sort options, status labels, column headings, pagination copy, bulk actions, and export messaging.

For billing and subscription products, it can create plan labels, billing cadence explanations, renewal messaging, usage-limit states, overage messages, trial copy, payment-failure messages, invoice actions, upgrade and downgrade confirmations, cancellation warnings, and reactivation states.

For AI products, it can create prompt guidance, file-context messaging, model-selection explanations, generation states, queue states, streaming messages, cancellation states, safety blocks, credit-limit messages, revision actions, result statuses, and history labels.

The skill also creates accessibility microcopy, including accessible names for icon controls, screen-reader announcements, status messages, instructions for custom controls, error associations, progress announcements, and hidden context that improves operation without duplicating visible content.

Localization guidance covers text expansion, pluralization, grammatical gender, date and number formatting, currency, right-to-left behavior, dynamic variables, line wrapping, sentence fragments, truncation, button widths, and content-token architecture.

It can produce reusable content tokens such as button.save_changes, error.network_retry, empty.projects_first_use, toast.invitation_sent, and dialog.delete_account.title.

The skill creates privacy-conscious analytics event maps for error exposure, error recovery, button interaction, empty-state activation, confirmation behavior, Undo usage, onboarding hints, loading cancellation, retry actions, permission confusion, and success-state completion.

It also creates microcopy audits, inconsistency reports, severity-ranked findings, experiment roadmaps, developer handoffs, component-state matrices, acceptance criteria, regression tests, QA checklists, risk assessments, and structured JSON specifications.

It prioritizes clarity, truthfulness, recovery, consistency, accessibility, localization readiness, verified system behavior, user dignity, and implementation feasibility.

The core commercial promise is: create interface language and interaction states that help users understand what is happening, what to do next, what went wrong, what was preserved, and how to recover.