Agent harness
Every component working together in one agent conversation.
69 components grouped by the job they do, covering agent status, conversation, review, data, charts, dates, forms, collaboration, product and workspace, designed and built by the DesignMe team.
A live agent conversation built only from the components below. Pick a starter and watch them work together.
Every component working together in one agent conversation.
Show what the agent is doing, how far along it is, and what it is working with.
Compact agent status that reshapes itself as the task moves along.
Progress through named steps with a choice of four loaders.
A timeline of the agent's plan, notes, sources and edits that unfolds as it works.
A quiet, collapsible trace of steps, reasoning, searches or code edits.
Shows what is filling the context window and lets you compact it.
Every file read, search, edit and command, logged as it happens.
Grouped tasks with live progress, retries and a compact view.
Live credit and token spend for a run, with a budget cap that pauses the agent.
A single notification with success, error, info and loading variants.
Agent notifications that stack neatly and fan out when you hover.
Several agents working in parallel, each with its own lane, progress and hand-offs.
Every model and tool step of a run on one time axis, with retries and cost.
Everything for asking, answering and replying, by text or by voice.
Answers that stream in with numbered sources, feedback and next steps.
Switch between agent threads and reply without leaving the panel.
Message input with sources, slash commands, model switching and dictation.
Dictate a request, watch the waveform, then review the transcript before sending.
Suggested tasks that fill the composer in one click.
Drop files for the agent and watch each one upload.
Regenerate an answer and page through every version.
Quick ratings that open into reasons when an answer misses.
Moments where a person checks, steers or signs off on the agent's work.
Multiple-choice questions the agent needs answered before it continues.
One suggested change with a confidence rating, alternatives and undo.
Pick a sentence and ask for a clearer, shorter or friendlier version.
Layout values the agent adjusts live, with a preview beside the inspector.
The agent asks for access, with scopes and a duration you choose.
Pre-launch checks the agent runs one by one, with fixes for anything that fails.
Proposed edits highlighted cell by cell, ready to accept or reject.
Sources, tables, metrics and code the agent reads or produces.
The documents the agent pulled in, filterable by type and pinnable.
Agent-written takeaways paired with a chart you can scrub week by week.
Tasks filtered by status, with rows that fold away smoothly.
Highlighted source with a switch to view the agent's changes.
Record properties you can edit in place, with tags, links and a way to add more.
Numbered citations that preview the exact source passage on hover.
Client list with tags, last contact and relationship strength, sortable in place.
Structured fields pulled from a document, each linked back to its source text.
Lightweight charts for dashboards and reports, drawn in plain SVG and HTML.
Trend with a comparison period, range switching and a hover readout.
Two series by month, grouped or stacked, with values on hover.
Share of a total, with segments and legend that highlight together.
A tick gauge that fills to the score, with a status chip, trend and date range.
Daily activity over recent weeks, shaded by volume.
Headline numbers with change and a sparkline, for the top of any dashboard.
Pickers for single dates and date ranges.
Month grid with keyboard support and a sliding month change.
Quick presets beside a calendar, with a live preview while you choose the end date.
A month view with colour-coded events and a summary card on click.
A scrollable week of meetings and deadlines with a live now line.
Buttons, fields, pickers and sign-in flows for any product.
Variants, sizes and every state from loading to done.
Labels with shortcuts on hover or focus, plus a richer explainer.
A search field with recent queries, live results and a keyboard shortcut.
A select with flags, search and keyboard navigation.
Provider buttons with a connecting and connected state.
Email and password sign-in with validation and a show-password toggle.
A project enquiry form with inline validation, chips, a counter and a sent state.
Card, wallet or bank checkout with live formatting, a discount code and an order summary.
Threads and inboxes for teams working alongside an agent.
A discussion thread with mentions, reactions and sorting.
An inbox with unread states, requests to approve and attachments.
Find people, set what they can do and send the invite.
Hover Like to open a reaction tray, then pick one and watch it land.
Shipped the new pricing page today. Thanks to everyone who reviewed it this week, the comparison table finally sits where people look for it.
Pricing, onboarding, setup flows and model comparisons for the rest of your product.
Three quiet plans with a billing toggle and a highlighted pick.
Setup steps with progress, one open at a time and a finish state.
A three step setup with a progress header and validation.
A spending cap with a striped progress bar and an inline editor.
Live conversion with flags, a swap, quick amounts and the rate trend.
One prompt, two models side by side, with speed, cost and a vote.
Navigation, search, settings and automations around the agent.
Collapsible navigation for spaces, projects and recent threads.
Jump to files, components and actions from the keyboard.
Autonomy level, capabilities and model, with a clear unsaved state.
A project tree that marks every file the agent adds, edits or removes.
Trigger, condition and branches, with a test run that lights the path.
What the agent remembers, with pin, forget and a master switch.
What people usually ask before they start using the library.