Agent plan preview
An AI agent or automation can show a proposed multi-step plan before execution.
298 entries grouped by the interaction problem they help solve.
An AI agent or automation can show a proposed multi-step plan before execution.
An agent or automation run has started and spans multiple steps, tools, gates, or side effects.
An AI agent or automation can create side effects that affect customers, money, access, production systems, legal status, public content, sensitive data, or external recipients.
A generated answer makes source-dependent claims but no source route is visible.
An AI product displays model certainty, extraction confidence, recommendation score, classifier probability, generated-answer confidence, retrieval rank, or risk score with precision the system cannot justify.
A user is introduced to an AI feature whose abilities, limits, data scope, uncertainty, or review needs are not obvious from the normal interface.
A generated AI output can influence compliance, customer communication, security, legal, finance, operations, code, policy, or other high-trust work.
Users need to create or edit automations that run later based on events, conditions, schedules, or record changes.
The user needs a back-and-forth assistant conversation with follow-up questions and answer refinement.
Users need to verify generated claims, summaries, recommendations, or extracted facts against source material.
Users must judge whether an AI prediction, classification, recommendation, extraction, risk score, or generated answer is reliable enough to use.
Users can identify wrong, unsupported, stale, unsafe, biased, irrelevant, or wrongly sourced AI output after it is generated.
Generated content is expected to be revised before it is copied, saved, sent, published, or applied.
Users need a person because AI, automation, self-service, or scripted support cannot resolve the situation safely or acceptably.
An AI agent, workflow, deployment, or automation is ready to perform a high-impact step and must pause for human authorization.
A model version, provider, lifecycle stage, availability, behavior, or replacement plan changes in a way users may need to understand or act on.
Users must write or revise an AI request before generation, analysis, transformation, or automation begins.
The AI surface supports open-ended requests and users need examples of useful, supported tasks.
A user needs another AI-generated answer for the same request or a visible recovery path after response failure.
A user's AI request has missing object, audience, timeframe, source, workspace, permission, output-depth, or action-target boundaries.
Users need answer-wide evidence coverage before trusting generated content.
Generated text or structured content can be read or monitored before completion.
An AI agent or automation calls tools, functions, APIs, retrieval systems, commands, or integrations.
Users need to catch up on recent collaboration, app, repository, channel, project, meeting, reminder, or task activity.
Users inspect and decide on proposed edits, tracked changes, code suggestions, file diffs, or document redlines.
Users need object-attached discussion without changing the primary object content directly.
Users need future updates from a specific object, thread, channel, space, repository, topic, saved view, or query.
Responsibility, context, or conversation control transfers to another person, team, queue, shift, AI agent, live agent, or system.
An authorized user needs to bring another person into a workspace, organization, project, channel, repository, team, tenant, or shared object.
Collaborators need real-time spatial awareness in a shared visual or text surface.
Composed content needs to target people or groups inline.
Users need to understand current or recent availability before messaging, calling, assigning, routing, or joining.
Users need quick, lightweight feedback on an existing message, comment, issue, document selection, or work item.
A user needs to share one object or a small set of selected objects with people, groups, domains, links, or device targets.
A topic or parent message can generate multiple reply branches that users need to follow over time.
Users need to create new photo or video media inside the task.
Users need spatial or ordered movement between clear sources and destinations.
A workflow has multiple focusable elements, dynamic content, or overlays that users traverse with keyboard or assistive technology.
A tactile pulse reinforces a direct manipulation, boundary, threshold, selection, successful action, recoverable warning, or physical game/control event.
Users repeat a known command often enough that memorized keyboard acceleration saves meaningful time.
Current device coordinates materially improve a task such as nearby search, route start, check-in, delivery estimate, safety sharing, field work, or support diagnostics.
A touch-first interface needs an optional shortcut to secondary item actions, selection mode, preview, or drag pickup.
A mobile user action can be saved locally and retried after network or app lifecycle conditions improve.
A top-of-list gesture is a platform-expected shortcut for checking freshness on a scrollable data surface.
Users need to decode a QR or barcode to pair a device, open a known destination, redeem a ticket, start payment, enroll an authenticator, join Wi-Fi, identify inventory, or import a short payload.
A product supports multiple viewport widths, device classes, orientations, split-screen sizes, or foldable postures.
Users repeatedly perform simple item-level actions on touch-first rows.
A touchscreen interaction needs deliberate design for gesture vocabulary, thresholds, feedback, target sizing, cancellation, and equivalent controls.
Users need hands-free control, accessibility speech input, or rapid spoken command activation.
A wearable surface can answer a status, next-step, or current-metric question in seconds.
Users need to inspect recorded user, admin, system, security, or integration events.
Users need to scan or manage scheduled events across dates, days, weeks, months, or resources.
Users browse peer objects where visual preview, compact identity, and quick first-pass selection matter.
Users need to browse rich object summaries with preview, summary text, metadata, status, and actions.
Users need to investigate an aggregate chart mark by moving to the next planned level of a hierarchy.
Users need to save, curate, share, revisit, sequence, or maintain explicit items as a meaningful set.
Users need to choose among similar offerings using multiple comparable attributes.
Users need to monitor several related metrics, exceptions, and analyses together.
Users repeatedly inspect or update many records in shared columns.
Users need to see trends, comparisons, rankings, distributions, relationships, composition, anomalies, or uncertainty.
A row has supplemental detail that is useful for some records but would make the default table or list too dense.
Users consume a continuing stream of updates, posts, stories, comments, media, or collaboration activity.
Work progresses through visible stages and the team uses the board to decide what to pull next.
Users need to scan and act on a collection of objects using compact summaries.
Users choose or inspect objects by where they are, how near they are, or what area they fall inside.
Users repeatedly choose items from a collection and work with a substantial detail destination.
Users repeatedly return to a specific presentation of a changing data set.
A longer page has a small set of key facts or next steps that users need early.
Records share comparable attributes that users need to scan in aligned columns.
A compact category, state, phase, freshness marker, priority, or count helps users scan a group of objects.
Users need a chronological record of important events for one object, case, order, application, or process.
Rows have real parent-child hierarchy and shared comparable columns.
Users need to compare, edit, inspect, or preview adjacent panes in one durable workspace.
A page has several related sections and users do not need all details visible at once.
An urgent current-workflow condition requires acknowledgement, cancellation, retry, sign-out, review, or another explicit response before continuation.
Mobile users need extra contextual detail or controls while staying oriented to a map, feed, route, media item, or selected object.
Use this anti-pattern entry to audit or reject timed carousels that cannot provide immediate pause, stop, focus, hover, reduced-motion, and interaction-stability behavior.
Objects or selections have secondary commands that experienced users expect near the target.
Users repeatedly inspect selected records, tickets, alerts, files, assets, tasks, comments, or locations while staying in a work surface.
One short supporting explanation belongs in place beside the content it clarifies.
Users need to inspect object detail, metadata, comments, history, or light actions while staying on the current page.
Use this anti-pattern to review temporary drawers, inspectors, side sheets, and mobile drawer layouts for missing close and return behavior.
A temporary task needs the whole viewport for capture, immersive preview, focused setup, scanning, or dense editing.
Users need to inspect object metadata before choosing whether to open a link, mention, file, row, issue, or reference.
Use this anti-pattern entry to audit toolbars, row actions, cards, command bars, mobile action sheets, and generated UIs with symbol-only controls.
A local object or toolbar has three to eight secondary commands.
A short task must interrupt normal page interaction but should return users to the same context afterward.
Use this anti-pattern to review informational dialogs, help popups, status popups, release-note prompts, marketing modals, and preview modals that interrupt work without a required decision.
A local control needs brief explanatory content or light editing while page context remains visible.
Users repeatedly inspect files, messages, PDFs, images, videos, audio, invoices, contracts, designs, or attachments from a list or grid.
The default task has a safe recommended path for most users.
A user starts a short context-bound task that needs more room than a dialog but should not become a full page.
A compact control needs a brief supplemental explanation.
Use this anti-pattern entry to review forms, checkout, applications, settings, data tables, eligibility flows, upload flows, and disabled actions where required information is hidden in a tooltip.
Auditing destructive dialogs, action sheets, menu items, command palette actions, delete receipts, account cancellation flows, permission revocation, bulk actions, and high-impact execution reviews.
Autosave failed, stalled, expired, or became uncertain while users still have meaningful local work.
The action is destructive, irreversible, costly, security-sensitive, privacy-affecting, or externally visible.
Use this anti-pattern entry to audit products that ask for confirmation on many routine actions or use identical prompts for different levels of risk.
A conflict has been detected and a user must choose, merge, save-copy, skip, or commit a result.
A user has initiated a destructive command that can permanently remove, revoke, reset, deactivate, or cancel something valuable.
Auditing disabled primary actions, toolbar commands, menu items, toggles, form controls, checkout actions, invite flows, publish controls, exports, destructive actions, and workflow steps.
Form submission can produce one or more errors.
Users have unsaved or pending changes that cannot be recovered if they leave.
Use this anti-pattern entry to audit products that show Undo, Restore, Revert, or rollback controls without proving exact reversal.
A primary lookup, verification, upload, payment, search, device feature, online-only step, or channel can fail while the user still needs to finish.
A richer UI control, script, API, media layer, device capability, or third-party widget improves the task but is not essential to the outcome.
A pull request, merge request, rebase, merge, or branch update has source-code conflicts that require developer decisions.
A denied user can request access, switch accounts, ask an admin, join a group, or use a safe fallback to continue.
Users perform reversible sequences of edits and can undo too far.
Deleted objects remain recoverable after the user leaves the original workflow.
A load, save, submit, upload, export, sync, payment, or background request failed for a transient or uncertain reason.
An authenticated session can expire because of inactivity, overall lifetime, assurance policy, or reauthentication requirement.
A severe action affects repository, project, workspace, account, production, security, billing, or organization-wide scope.
The action is common and mistakes are likely.
A close, cancel, switch, or edit-mode exit would discard unsaved local changes.
Use this anti-pattern entry to audit forms, editors, imports, checkouts, onboarding flows, and configuration screens where validation can erase recoverable user work.
Users edit documents, files, pages, design files, records, configurations, or published content over time.
A current task has a time-sensitive warning, error, or important status change.
The message applies across several pages, routes, sections, records, or sessions.
A transaction or service journey has ended and users need durable proof plus next-step information.
Concurrent users, devices, branches, or background jobs changed the same object or location.
The service sets non-essential cookies or similar device storage technologies.
Auditing generated UIs, dashboards, tables, lists, or panels for blank states that strand users.
Use this anti-pattern entry to audit disabled submit, continue, create, save, delete, publish, invite, export, payment, and setup actions.
The product area can legitimately contain no user data.
A system or task failure blocks expected content or action.
Use this anti-pattern entry to audit loading, saving, syncing, uploading, report generation, billing retrieval, AI generation, and import flows that can hang.
A visible object or section has local status, warning, success, or next-step information.
The content shape is predictable.
A short action, request, save, submit, refresh, sync, or fetch is actively processing and progress cannot be meaningfully measured.
A current value exists inside a meaningful known range.
A message should be encountered before the page H1 but is not the page's main task content.
Users receive multiple asynchronous updates across objects, jobs, collaborators, approvals, or reminders.
Connection loss or server reachability changes the user's current task.
A route, URL, public page, file, object link, or deep link cannot be found.
A signed-in user lacks permission to view, edit, publish, export, delete, approve, share, administer, or configure a resource.
A service is in alpha, private beta, public beta, or another explicit pre-live phase.
A system operation has a measurable total or bounded progress value.
A whole service, transaction entry point, or route is closed or not ready to handle requests.
Urgent public, safety, operating-status, outage, closure, or availability information applies to the whole site or service.
A workflow has several named stages and users need to understand current, completed, pending, blocked, failed, waiting, or skipped status.
A user action has completed and users need proof, object identity, or next-step orientation.
Local changes, files, messages, uploads, or records need to reconcile with a remote service or another device.
Use for short non-blocking confirmation after explicit actions such as save, copy, send, archive, invite, or queue export.
Use this anti-pattern entry to identify and replace transient-only handling of payment, save, permission, account, upload, deletion, and security failures.
Use this anti-pattern entry to identify transaction success handled only by transient feedback.
A user must understand a serious consequence before taking or skipping an action.
The product needs a complete postal, delivery, billing, service, or official contact address.
The list is long but values are known.
Users spend meaningful time entering form content or application progress.
The product needs bank account details to send money, collect money, set up payroll, save payout details, issue refunds, or initiate bank-to-bank payments.
Users need to create, update, or merge many structured records from CSV, XLSX, TSV, copied spreadsheet data, or another system export.
A legal, editorial, messaging, moderation, or display constraint imposes a clear maximum length.
A product configuration has many related fields across named sections.
A selected radio, checkbox, or select option needs one short related follow-up field.
Users already know the exact date or can look it up easily.
Users are scheduling, booking, planning, or selecting a recent or future date.
Users need to select a reporting, analytics, booking, scheduling, campaign, or eligibility period.
A downstream field's valid values or requirement depend on a previous answer.
Users create or edit objects that should not become visible, active, submitted, or externally effective until a later action.
Users often upload files from a desktop file manager and benefit from dropping one or more files into a visible target.
The product must capture an email address for contact, receipt, notification, invitation, account identifier, recovery, or verification.
The service needs a document, image, scan, spreadsheet, media file, or other original file.
Users frequently update one visible value or one row property.
A single field has a specific correctable problem.
The value has a stable character pattern and users benefit from live grouping.
The value has a fixed currency, unit, percentage, domain, or commonly understood icon.
A transaction spans several ordered pages or sections.
The product must capture a person's name for display, correspondence, booking, application, payment, account details, identity matching, support, or official records.
The user is creating an account with a reusable password.
The user needs to enter an existing password, passphrase, or PIN-like memorized secret.
Users need to pay with a credit, debit, prepaid, or network card.
The product must capture a phone number for calls, texts, support callbacks, appointment changes, recovery, delivery coordination, verification, or contact records.
Use this anti-pattern entry to audit forms where conditional logic, validation, saved drafts, or server schemas might require fields that are not visible or reachable.
A user has provided multiple answers and should verify them before a consequential submit action.
A compact set of related fields should be reviewed together before one submit.
One answer or decision should be read and completed without competing form questions.
The answer is short and user-authored.
Users need to write sentences, paragraphs, comments, explanations, notes, or descriptions.
Users choose appointment, meeting, delivery, pickup, reminder, or service times.
A setup, import, object creation, or configuration task has strict sequencing.
Users repeatedly choose similar values and the product can explain a likely starting value from current context or prior behavior.
Users have different monitoring priorities and need a persistent personal or role-based dashboard arrangement.
Users need to mark preferred or personally important objects for later return.
The product or site offers current content, task flows, or interface text in more than one language.
Users receive enough notifications that they need control over type, channel, device, frequency, timing, or source.
Users need stable quick access to a small set of known high-priority objects.
Users need to revisit and change communication, consent, topic, personalization, privacy, channel, language, or data-sharing choices.
Users repeatedly operate a broad set of tools, files, commands, apps, workspaces, or tasks.
Users are working in a record, case, conversation, or workflow where choosing the next action is costly or error-prone.
A product has multiple settings sections and users need a durable destination to find the right area.
Users need to switch between scanning many records and reading or editing with more space.
Users work with multiple panes or regions around one active task and need different arrangements for comparison, editing, reading, or focus.
Expert or repeat users need precise retrieval across fields, operators, phrases, dates, ranges, exclusions, functions, or prior search history.
Users know a word, name, or identifier.
Users are exploring a broad catalog, service directory, help center, learning library, product collection, or content repository.
The data set has structured attributes users understand.
A few common filters should stay visible and directly toggleable near the content.
Users need to narrow the current search results, browse results, table, card grid, or list by multiple criteria.
Use this anti-pattern to review search result pages, list filters, dashboard filters, saved filters, saved searches, mobile filter drawers, no-results recovery, and generated clear/reset actions.
Use this anti-pattern to review feeds, search results, product grids, media streams, activity streams, catalogs, and generated infinite-scroll pages.
Search, browse, or filter controls can produce an empty result set.
Users submit free-form searches where spelling and typing mistakes commonly block useful results.
Users revisit a search surface and benefit from rerunning recent query wording.
Users need discovery help in a large item, product, content, service, or action space.
Users repeatedly need the same filter criteria on a list, table, board, queue, base view, or dashboard.
Users repeat the same search criteria across sessions or operational cycles.
Search activity persists beyond the current session and users need to inspect or control it.
Results contain enough text that users need to scan for where the query matched.
The same query can search different repositories, sites, workspaces, spaces, channels, result types, teams, or organization-wide indexes.
The system can predict likely queries.
Users need to compare or act on a result set in different orders.
Users search large datasets and benefit from seeing result hits before full submit.
A resource, row, card, selected set, or local page object has several contextual commands.
A user-initiated action has a few immediate valid outcomes.
A form, dialog, panel, card, or workflow footer needs two to four related visible commands.
Users can choose zero, one, or many options.
Users answer a lightweight choice question from a small set of short labels.
Users choose one value from a large, dynamic, or filterable option set.
Use this anti-pattern entry to audit custom dropdowns, styled select replacements, portal menus, command-palette-like selectors, and generated UIs that imitate native controls.
Users choose one option from a visible static list.
A complex application or editor has many stable commands organized into known groups.
Users choose many values from a long predefined list.
Users assign, link, reference, invite, route, or attach an existing entity.
Users must choose exactly one option.
Users need a lower and upper bound on one numeric scale.
There are a few mutually exclusive modes.
The user chooses one option from a moderate known list.
Users adjust one bounded value such as brightness, volume, opacity, zoom, intensity, cost tolerance, or preference strength.
Users adjust a small bounded count such as guests, copies, items, seats, nights, minutes, retries, or quantity.
A single setting, feature, preference, or hardware-like control has two clear states.
Users build a selected set by moving items from an available collection to a selected collection.
Users choose one or more values from a meaningful hierarchy.
Users need to save and resume drafts, regularly update their data, manage records, check ongoing status, collaborate, or protect repeated access.
A submitted request needs authorized approval before a deployment, purchase, access grant, publication, content change, policy exception, financial action, or workflow step can proceed.
A user must set, change, or clear who is accountable for a work item.
Users reserve an offered appointment, service, class, room, resource, seat, table, pickup window, or visit.
A user is ready to purchase selected items, services, tickets, subscriptions, donations, or orderable options.
A transaction has several meaningful sections or tasks and progress is saved.
A user must prove access to a mailbox before account activation, invitation acceptance, recovery eligibility, or sensitive notifications.
A captured phone number must be proven reachable before activation, contact-route reliance, recovery eligibility, phone-number change, or a sensitive action continues.
Users need to create a visible profile for collaboration, community, directory, marketplace, public contribution, support, or social interaction.
Users have submitted credentials or authenticator output and need a safe verification result.
New users need orientation, setup, personalization, or instruction before the regular interface can deliver value.
Users need to recover an account because the existing password is forgotten, suspected compromised, or unusable.
The user must pay a known fee, invoice, balance, application charge, renewal, subscription, donation, deposit, or service amount.
Users already have an account or profile and need guided help to complete, review, enrich, or update it.
Users need to make content, products, pages, releases, site changes, campaigns, or configuration visible to an audience.
A team or individual repeatedly reviews many independently queued items.
Users coordinate time across people, resources, calendars, availability, or recurrence.
Users need to inspect and change persistent app, account, workspace, privacy, notification, display, integration, device, or system behavior.
Users need to access an existing account or protected destination.
A user is about to start one named service, transaction, booking, application, check, request, or registration.
A page needs to display a known set of tasks with status and links to start or resume them.
A view, editor, canvas, media surface, table, or selection needs repeated related commands close to the work area.
A sign-in, new device, risk signal, or sensitive action requires proof beyond the primary credential.
Age or age band controls whether users may access content, features, commerce, community interaction, personalization, or data collection.
The product must support security investigations, compliance review, legal discovery, privileged-action review, or enterprise governance.
Users need to reduce exposure to a person, account, conversation, channel, app, bot, word, topic, or thread.
The product needs a user's active agreement for optional data use, marketing, research participation, personalization, partner sharing, AI training, or sensitive-data processing.
The product uses cookies, local storage, pixels, tags, SDKs, or similar technologies that are not strictly necessary.
A user, agent, automation, or admin tool is about to execute a high-impact action that can affect money, access, production systems, legal/compliance state, customers, external recipients, sensitive data, or safety.
A UX review, privacy review, legal review, accessibility review, or support report identifies manipulation in a consent, cookie, marketing, sharing, AI training, or data-use choice.
Users need to download or transfer a copy of account, workspace, personal, product, activity, or organization data.
Users can create an account and need a self-serve or direct request path to close it and delete associated data.
The user may be harmed if someone nearby sees them viewing or using the current sensitive service.
Auditing a product that supports account creation and may be hiding, obstructing, disguising, or underspecifying account deletion.
A user must accept terms of service, conditions of sale, policy documents, service agreements, acceptable-use rules, or legal disclosures before access or transaction completion.
The service is eligible to identify itself as an official government or public-authority website.
A product currently asks for device, browser, or app permissions before the relevant feature context exists.
A feature needs operating-system, browser, or device authorization to access location, camera, microphone, photos, contacts, notifications, Bluetooth, clipboard, motion sensors, storage access, or another powerful feature.
Owners or admins need to manage durable access to spaces, sites, repositories, projects, folders, datasets, boards, environments, or sensitive objects.
Users need ongoing control over personal data collection, saved activity, visibility, app access, device permissions, connected services, data sharing, or personalization.
Users need to flag content, behavior, an account, conversation, listing, repository, ad, or message for policy, abuse, spam, privacy, legal, or safety review.
A threat signal indicates phishing, malware, deceptive site, unsafe download, invalid certificate, insecure connection, mixed-content submission, suspicious redirect, file preview risk, or account-security danger.
Users need to inspect, copy, verify, rotate, transcribe, or compare a sensitive value that should normally stay masked or redacted.
An authenticated session has expired or been terminated while the user was on a protected task.