Whole pages, composed only from the components in this registry and shelved by the dashboard they belong to. Installing one drops a route into your app — the page, its parts, and the sample data it reads — with nothing to import from a package. Every one runs here: open a quick look and drive it.
109 pages
A store's back office: sales, the order book, products, customers, stock, the parcels still to go out and the till in the shop.
A shop's own dashboard: gross sales, orders, new customers and refunds against the month before, revenue by category as a ring beside its numbers, the funnel down to an order, what moved most, and every order in the window filtered from the URL.
A sales page: bookings, open pipeline and win rate, the quarter's quota as a goal tracker, booked against fulfilled value by month, the pipeline by stage, the rep leaderboard, and every deal on the board.
An order operations page: four KPIs read against their own targets on bullet tracks, the week's volume by weekday, booked value by region, where the whole book stands as one bar, and every past order in a filterable table.
A server-paged orders table with status tabs, a payment filter and search, and an inspector for the order you click: its lines, its payment, its timeline, and a refund that can be refused.
One order's page: its lines from the catalogue at the prices it charged, the parcel's stages, the customer with their face and their other orders, how it was paid — and a refund and a fulfilment that each say why when they cannot run.
The catalogue as a table: search by name or SKU, facets on category and status, stock drawn against each reorder point, units sold in the last thirty days, a row menu that archives, and a bulk bar for many rows at once.
One product's page: its price and stock, twelve weeks of units sold against their own average, the orders that hold it with each customer's face, and a form that edits the price and the stock — and refuses a negative.
A server-paged accounts table: status tabs, search, a plan filter, sortable columns, a column menu, an export, and bulk actions over the rows you tick.
One account's page: a record header with its status and owner, and five sections — its facts and revenue, its orders, its invoices, its history, and an editable note.
A stockroom page: what is on the shelf and what it is worth, a low-stock banner, the SKUs at or below their reorder point with the quantity that brings them back, inventory turnover by category, and the suppliers holding the most stock.
A warehouse floor: parcels in transit, delivered this week, open exceptions and mean transit time; throughput by service at two bucket sizes; a stage bar for every parcel that has stood still longest; what each carrier is holding; and the queue of parcels that stopped.
A till for the shop: the catalogue as drawn tiles with what the shelf holds, a ticket with a customer, named steppers, a discount code, 8% tax and cash, card or voucher, and a receipt that prints alone.
The numbers a subscription business watches: revenue, usage, retention, where the users are and what is happening right now.
A product analytics dashboard: sidebar shell, four headline metrics with sparklines, traffic over time, sources, top pages, team activity, and the newest signups.
A revenue page: MRR with its month and year deltas, the ratios beside it, twelve months by plan, the quarter's target, where the month's movement came from, the largest accounts, and how much of what customers pay for they use.
A product analytics page: daily, weekly and monthly active users, stickiness, a feature adoption heatmap, a cohort retention curve, the accounts with the most people in the product, and the event stream behind them.
A retention page: signups and the share still active a week, a month and two months on, the cohort heatmap behind those numbers, the same grid as curves, which plans the survivors sit on, and every cohort's size against its target.
A worldwide traffic page: sessions, countries, weighted p95 and booked revenue, a dotted world map with a marker per country, the busiest countries as share bars, revenue by region, and the newest orders from anywhere.
A live operations page: four ticking numbers with sparklines behind them, a scrolling country strip, a feed that fills itself, sessions by country, and p95 by country against the objective.
A recurring-revenue page: MRR and who is on the book, a year of movements as a waterfall that closes exactly, the plan mix, retention by signup cohort, the trials about to run out, and every subscription ever opened.
A customer acquisition page: how many accounts there are and what they are worth, the funnel from ad clicks down to paying workspaces, signups a month with the month in progress projected out, campaign targets, and the whole directory.
Deals, leads and campaigns: the pipeline, who is in it, and what it cost to fill.
A CRM page: the quarter's target as a ring, customers, open and won deals, leads by source, the week's follow-ups as a checklist, the pipeline stage by stage, and every lead in a table that converts one to a deal.
Deals as a board by stage, each column summed in money under its title, with a drawer that prints the deal a reader opens.
Every lead with their face, company, source, status, owner and value — facets on source and status, four figures that recount as you work, and a row menu that converts a lead into a qualified deal on the pipeline.
A campaign dashboard: spend, conversions, cost per acquisition and return on ad spend, daily spend against conversions, the channel mix, the biggest campaigns, and the UTM links behind them.
The campaign slate: spend by channel stacked month by month, then every campaign with its spend, clicks, conversions, cost per conversion and conversion rate — a channel facet, and Pause or Resume from the row menu of a campaign that has not ended.
The help desk: the queue, the inbox, the help centre and what changed since a customer last looked.
A help desk page: open tickets and median reply times, volume opened against resolved day by day, SLA attainment on a gauge, customer satisfaction on a score ring, the outstanding queue, and the agents behind it.
A two-pane inbox: threads with unread markers and priority on the left, the conversation, its assignee and a reply composer on the right.
A searchable help centre: one card per shelf with its own count, a search box that looks through the whole article rather than its title, articles that open in place, and a way to reach a person when none of them answers.
Every release in one column: the version, the day it went out, the kinds of change it carried, and its lines — with a subscribe form that answers through a server action.
Deploys, uptime, incidents, alerts and logs — what is broken, since when, and who has it — and the public status page.
A delivery page: deploy frequency, lead time, change failure rate and time to restore, production deploys a week, the last releases on a timeline, every environment's recent deployments, and the pull requests still waiting for review.
An operations page: service status, ninety days of uptime, latency percentiles with one overnight spike, cluster load, errors by hour and weekday, the regions traffic is served from, open incidents, and the gateway log.
The console the responders work in: every incident with severity and status filters, the selected one's timeline beside the list, who is holding the pager, and a declare-incident dialog that writes a real row.
Every monitoring rule the workspace watches: what is firing as callouts at the top, the rules table under them with severity, state and the channel each one pages, and a silence action with a duration.
A tail of the platform's log with level and service filters, the saved queries a log is actually read through, a live dot the reader controls, and an export of exactly the lines on screen.
The status page customers read during an outage: one banner for the whole platform, every region under it, ninety days of uptime per service, what is still open, and a fortnight of incident history with the updates as they were published.
The people team's desk: the directory, roles, invitations, the leaderboard and who is out this month.
A people page: headcount, attendance, the leave queue and the average KPI, full-time against contract, the week's attendance as a stack, a work calendar, every employee in a table with a CSV export, the requests to approve or decline, the departments and an arrival heatmap.
Everyone with a seat in a workspace: how many people against how many the plan covers, a directory ordered by role, name or activity, an invite dialog, and removal behind a confirmation.
Roles down the side, resources across the top, one checkbox per pair: a panel spelling out what a ticked row actually grants, and a save bar that appears the moment a box changes.
Seats that have been offered: what is still waiting with the time each link has left, resend and revoke on every one of them, and a table of everything already settled.
Who is ahead: a podium for the top three, a ranked list of everyone else, and one switch each for the window and the measure — revenue booked, deals closed, or support tickets answered.
Every leave request with the employee's face, its kind, dates and days, facets on status and kind, Approve and Decline on each pending row, and the month as a strip of days counting who is out on each — pick a day to see who.
Cash in and cash out: the ledger, invoices, payments, balances in four currencies and a treasury held in crypto.
A finance page: cash, net burn, runway and margin, the month's profit and loss as a waterfall, where the spend went as a treemap, everything still owed by how late it is, and the payouts leaving next.
A server-paged invoices table under four aging buckets and an overdue callout, with send and void actions that move an invoice between buckets and can be refused.
Every payment the store took, sent back or failed to take, read off the order book: the month so far in three figures against the same days of the month before, a table with each customer's face and the card behind the charge, method and status facets, and a CSV export written by the server.
The store's money by currency: a card per balance with what is available and pending and a pill on thirty days ago, a verification callout, the latest and upcoming transactions with a face on every person, an exchange-rate chart with period chips, and a converter that works as you type.
The wallets: an overview of every holding added up, each wallet on one line with its share of the balance, a buy-and-sell desk that converts the amount to dollars as it is typed, the ledger newest first, the three flows by month, and every holding in a table.
The work and the team around it: tasks, a timeline, to-dos, notes, a calendar, the automations that run it, chat, mail, the team feed and files.
A project office: what is in flight and on track, hours logged and overdue tasks, points planned against done at three ranges, the team and its rates, the week's deadlines, delivery by quarter, effort by column, and every project in a table that opens the roadmap or archives one.
A team's work as a kanban board with WIP limits, filters by owner and priority, a list and a data table of the same rows, and an inline add.
Every open project as a row of bars over the calendar, with milestones, a weeks or months zoom and a panel for the bar you pick.
The team's tasks as a list: tabs by state with counts, a search, a filter popover, cards with a tick, a star, the owner, the due date and the subtasks, and a floating add.
A board of notes — prose and checklists, labelled and pinned — with a rail of folders and labels, a search, a grid or a list, and an add dialog.
A month grid built from the calendar entries, an agenda for the dates you pick, and a panel that inspects any entry — attendees, location and reference included.
Automations as node canvases: every workflow with its state and last run, the open one drawn as steps joined by arrowed curves on a plane that pans and zooms, an inspector that renames a step, a switch, Run now and the run history.
A two-pane chat: direct messages and groups with presence, receipts and unread counts on the left, the thread by day with a composer on the right, and the open chat in the URL.
A three-pane mail client: folders with unread counts and four colour-dotted labels, a searchable list with All and Unread, and a reader with reply, archive, delete, star and mark unread — one pane at a time on a phone, and a compose sheet.
The team feed: a composer that attaches a poll, a link card or a product, posts whose media are drawn rather than photographed, likes, comments that open in place, and a rail with your profile, who to follow and what is trending.
The drive at a glance: documents, images, videos and the rest in four toned tiles with their share of what is used, the storage band against the plan's quota, transfers and uploads by month, and the files touched most recently.
The drive one folder at a time: the open folder in the URL, every row selectable, a star on each, rename and delete from the row menu, an upload sheet with a dropzone, and a bulk bar for what is ticked.
A hotel's front desk: tonight's arrivals and departures, the bookings, the room rack and the guests.
Tonight's house: occupancy, ADR and RevPAR, an availability calendar, booking channels by week, the three guest lists and an arrivals sheet.
The whole book, server-paged: the next thirty nights, new business and cancellations in four figures, the book from today, the past or all of it, a status and a room-type facet, and a menu on every row that checks a guest in, checks one out or cancels a stay.
The house tonight as a floor plan: thirty-six rooms on three floors, each tile drawn to its room and filled by tonight's state, with a legend that counts them, each floor's occupancy, a sheet per room that checks its guest in or out, the day's turnovers and the rack rates against what tonight pays.
Every guest on the book, worked out of the bookings under their name — stays, nights, spend, last and next stay — biggest spenders first, with search, four figures and a sheet of every booking a guest has made.
A clinic's day: the appointments, the patients, the doctors and how full the schedule is.
A clinic's month: appointments, new patients, procedures and revenue in four toned tiles, visits by department, the calendar with each day's book, the desk's notes, the top procedures, and everything booked from today with a check-in on every row.
Every visit in the clinic's book: the patient with a face, the doctor, the department, when and for how long, the procedure and where it stands — from today, the past or all of it, with a facet on department, doctor and status, four figures, and a check-in or a cancellation on every visit still booked.
Every patient in the book, worked out of the visits under their name — visits, first and last, the departments that saw them and the next appointment — with this month's new patients in the figures, search, and a sheet with each patient's history.
Every doctor on the schedule as a card: the department, the hours booked in the last twenty-eight days and their share of the clinic's, the month's visits seen and cancelled, the next opening on their next clinic day and who is next through the door — narrowed by department.
An agency's book: the listings, the agents holding them and the viewings booked this month.
An estate agency's book: active listings, the month's enquiries, the quarter's closings and the commission on them, closings against viewings by week, month or year, the most-enquired listing in a spotlight, closed against in progress, the next viewings, the agents, sales by channel, the portfolio and a table of what is on the market beside a calendar of viewings.
Every listing on the book — kind, city, price, rooms and area, status, the agent with a face, views and enquiries — with the book added up in four figures, facets on kind, status and city, and a switch that draws the same page as tiles the kit draws in place of photographs.
One listing, drawn with the kit rather than photographed: the facts with the price per m² against its city, the interest it has drawn week by week, its price history, the agent with a face and a number to call, the next viewing, and the deal — Mark under offer and Mark sold through server actions that refuse a listing already off the market.
Every agent holding a listing, a card each with a face and a number to call: what they have on the market, what they sold and let this year, the enquiries their book drew with twelve weeks of them as a sparkline, and their next viewing — ordered by closings, enquiries or name.
The next four weeks of viewings, day by day — the time, the listing, where it is and the agent showing it with a face — with a calendar that narrows the agenda to one day, four figures, and Reschedule through a server action that refuses a slot that has passed.
An online school: the courses, the students on them and how far each has got.
A learner's home: a welcome card with their four numbers, two learning paths, a leaderboard, the cohort's success rate, the catalogue by month, and every course with a Continue button that writes back.
Every course in the catalogue — category, level, the instructor with a face, lessons, learners enrolled, the share who finish and the rating — with the catalogue added up in four figures, facets on category and level, and each course opening its own page.
One course: the header with its level, lessons and instructor, four figures, twelve weeks of enrolment, the students on it with their progress, the instructor with a face, the completion funnel from enrolled to finished, and every lesson numbered with how many students are past it.
Every student with a face — the courses they hold, their average progress, when they were last in and where they stand — with the roster in four figures, facets on status and course, and a sheet per student listing their courses and enrolling them on another.
A product with a model in it: tokens and cost, two assistants, an image studio and the voice jobs.
What the product spends on hosted models: spend, tokens, P50 and P95 latency by day, the provider split, the status codes, and a trace log with an inspector.
A two-pane assistant: threads by day in a rail, a greeting with quick prompts, and a transcript whose replies come from a seeded composer.
The assistant as one centred column: a greeting by name and time of day, six suggestion cards that fill the composer, a composer with files and a model, and answers with lists, code and a copy, regenerate and rating row.
A settings card and a two-column gallery of generated pictures — each an abstract drawn from the row's seed and palette, never a photograph.
Text to speech: a script with a counter, five voices as selectable rows, speed, pitch and stability sliders, and a history whose rows play as a stepping waveform.
The screens every dashboard needs and none owns: sign-in, onboarding, settings, billing, a profile and the error states.
A six-section settings page: workspace and brand, members, billing and usage, API keys, notification preferences, and a danger zone — with a save bar that appears the moment anything changes.
An append-only trail read either as a rail or as a table, filtered by actor and resource, with a diff viewer for what each entry changed.
The first page of a new workspace: a setup checklist whose steps are answered from the data, three counts of how far it already is, and a card for each step still to do.
A four-step wizard — workspace, invitations, sources, done — that keeps the step it is on in the URL, validates each step before it moves, and finishes through a server action.
A 404 inside the shell: the address that was asked for, a search over everywhere the product goes, and the last few things you changed to pick up from.
The two ways a page fails: a 500 with the details behind a toggle, the reference, the incident already open and a way to try again, and a scheduled-maintenance notice with the window and what it takes down.
A 403 inside the shell: what is restricted and why, who can grant it, and a request-access form that validates through a server action and confirms in place.
A pricing page for the workspace you are in: the tier it is billed on with its seats filled, every other tier priced monthly or yearly, and a dialog that confirms the change.
A billing history for every workspace an organisation runs: what is outstanding, what has been paid, one row per invoice with its download control, and the wallet the next invoice is charged to.
Metered usage for the term a subscription is in: seats, requests and tokens against what the tier includes, an overage forecast against the ceiling, and where the organisation's usage goes workspace by workspace.
A centered sign-in card: email and password with inline errors tied to the field that failed, a remember switch, links to recovery and signup, and single sign-on placeholders.
A split signup page: name, email, and a password with a live strength meter beside a brand panel of highlights, and consent gated before the account is ever created.
A centered recovery card: one address field, and a check-your-inbox confirmation that never reveals whether the account exists.
A centered card for setting a new password from an emailed link: the password twice, a mismatch that marks both fields, and a confirmation that ends on the sign-in page.
A centered card for confirming an email: a six-digit one-time code in two groups of three, a resend button held shut by a 30-second countdown, and a verified confirmation.
The signed-in person's own details: avatar, name, email and timezone in one saved form, with the account facts the form cannot change beside it.
Account security in three panels: change the password, pair an authenticator app with a six-digit code, and sign out of the places this account is signed in from.
A channel by event matrix — email, in-app and push against every kind of notification the workspace sends — with the digest window under it.
Every key in the workspace, live and revoked, with a create dialog that shows the whole key once and a confirmation behind every revoke.
The connector catalogue as cards — connect, disconnect, or retry one that is refusing — filtered by the categories the catalogue actually uses.
A four-step importer: drop a CSV, map each of its columns to a field, look at the first rows as the mapping reads them, and write the rest — with every refused row named by its line number.
Turn on a second factor: a QR of the otpauth URL with the same secret printed beside it to type, a six-digit verify, and the recovery codes handed over once it is on.
Sign in without a password: one address in, a sent screen that names where the link went without printing the address back, and a resend held shut behind a countdown.
Sign in with a passkey, or make one first: both halves of the ceremony mocked behind server actions, with a refusal under the button that produced it and three sentences saying what a passkey actually is.
Take a seat in a workspace: who invited you, into what and as what, a name and a password to set — and its own page, with its own heading, when the token has expired, been withdrawn, been used, or is not one we know.
The signed-in member's page: a tinted cover, tabs kept in the URL, the audit trail as a stream, connections, teams and a projects table.
Every notification as a page: an unread count in the header, search, status and kind filters, inline accept and decline on an assignment, and a paged list over db.notifications.
A workspace at a glance: revenue, new subscriptions, paid orders and sessions this hour with their changes, revenue by week against the weeks before, subscriptions by month, the team's roles, a support thread, sessions this week against last, the latest payments and a payment link.
A teammate's profile without a cover: the face and the name with Follow and Message, four figures, about, skills with a meter each, the teams they are on with the faces in them, a contribution heatmap over six months, and the latest of what they did.
One command, and the page — its parts, its sample data, its server actions — is source in your repository. No account, no license key, nothing to keep in step with a release.
npx shadcn@latest add @vibra/saas-overview