wireframe skill
Produce low-fidelity black-and-white UI wireframes as SVGs or viewer pages. Use when asked to wireframe, sketch a screen, draft a layout, make a low-fi mockup, or publish wireframes.
Is the wireframe skill safe?
Clean: nothing in its files matched our rules. We read 5 files in the folder on 2026-10-12.
No findings.
Install the wireframe skill
A skill is a folder. Copy it into your agent's skills folder and the agent loads it when the task matches its description.
git clone --depth 1 https://github.com/paperclipai/paperclip.git /tmp/paperclip mkdir -p ~/.claude/skills cp -r /tmp/paperclip/packages/skills-catalog/catalog/bundled/product/wireframe ~/.claude/skills/wireframe
In the Claude apps, zip the folder and upload it from the Skills settings. The folder on GitHub
The instructions your agent would load
SKILL.md as published, without the frontmatter. Read it on GitHub
Wireframe
Produce low-fidelity, black-and-white UI wireframes as standalone SVG files. The goal is to communicate structure — what goes where, in what order, at roughly what size — without committing to colour, brand, or polish.
When to use
Trigger on phrases like:
- "wireframe a [screen / page / flow] for X"
- "low-fi / lo-fi mockup of X"
- "draft a layout for X"
- "rough sketch of the [dashboard / settings / login / ...] page"
- "show me how X would lay out before I build it"
Skip and defer to frontend-design (or similar) when the request mentions: brand, polish, real components, "production-ready", colour palettes, hi-fi, Figma export, or actual code/HTML/React deliverables.
House style — non-negotiable
Wireframes are diagnostic, not decorative. Lock these tokens on every output:
If you need to highlight a specific region for a callout, use the annotation layer (red dashed). Never colourise the wireframe itself.
Workflow
- Confirm scope. What screen(s)? Which viewport (desktop / tablet / mobile)? Single screen or multi-screen flow? If unclear, ask one question, then proceed with the most likely default.
- Pick a canvas from the table above. State the viewport in your reply.
- Compose from primitives. Read references/components.md and assemble the screen from the primitive snippets. Snap every coordinate to 8px.
- Write the SVG to a file. Default path: wireframes/.svg in the working directory. Filename slug describes the screen (login.svg, dashboard.svg, settings-account.svg).
- Emit a textual annotation list in your reply, mapping each numbered region in the SVG to a one-line description ("1 — primary nav, 2 — search input, 3 — list of recent items"). This makes the wireframe accessible, queryable, and reviewable in text-only channels.
- For multi-screen flows, produce one SVG per screen and a summary flow.svg that arranges thumbnails left-to-right with arrows between them.
Quick start — minimal SVG
<svg xmlns="http://www.w3.org/2000/svg" width="1280" height="800" viewBox="0 0 1280 800"
font-family="-apple-system, system-ui, sans-serif" fill="#fff" stroke="#000" stroke-width="1.5">
<!-- canvas border -->
<rect x="0" y="0" width="1280" height="800" />
<!-- example: a button -->
<g transform="translate(48, 48)">
<rect width="120" height="40" rx="4" />
<text x="60" y="25" font-size="14" text-anchor="middle" stroke="none" fill="#000">Continue</text>
</g>
</svg>Two house-style gotchas worth memorising:
- Always set stroke="none" on elements (text inherits the parent stroke and gets a halo otherwise).
- Always wrap text fill explicitly (fill="#000") since the parent group fill is #fff for boxes.
Primitive library
The full set of reusable primitives lives in references/components.md. Load it whenever you need a primitive whose exact markup you do not have in working memory. Do not re-derive primitives from scratch — copy the snippet and adjust coordinates.
Primitives provided:
- Inputs: button (filled, outlined, icon), text input, textarea, dropdown, checkbox, radio, toggle, search input
- Layout: card, section divider, sidebar, two-column, three-column
- Navigation: navbar, tab bar, breadcrumb, pagination, sidebar nav
- Content: heading, paragraph block, list row, table, key-value pair, metric tile
- Media: image placeholder, avatar (circle/square), video placeholder
- Overlay: modal, drawer, toast, tooltip, dropdown menu (open state)
- Annotation: numbered callout, dashed region highlight, arrow connector
Grid, palette, and type scale
For exact pixel values, palette tokens, and type sizes, see references/grid-system.md.
Worked examples
references/examples.md contains four complete wireframes you can copy and adapt:
- Login screen (mobile, 375×812)
- Admin dashboard (desktop, 1280×800)
- Settings page with form (desktop, 1280×800)
- Modal confirmation overlay (desktop, 1280×800)
When the user's request is close to one of these, start from the example and modify, rather than building from blank.
Output convention
Every wireframe response should include:
- The SVG file written to disk (path stated explicitly).
- The SVG inlined in your reply (so it renders in markdown previews).
- A short numbered annotation list mapping regions to intent.
- Any explicit assumption you made (viewport, signed-in state, empty/populated, dark/light not applicable since monochrome).
If the user requests a website / viewer page
Trigger on phrases like "make a page that shows the screens", "single page I can scroll", "build a viewer", "let me click through the wireframes", "show them all on one page", or any request to bundle multiple wireframes into a browsable artifact (not a production site).
Build one static index.html that loads the SVG wireframes directly. Do not turn this into a React app or component library — it's a review surface, not product UI.
File layout (default):
design/<task-slug>/
index.html
wireframes/ # the SVGs from this skill
screenshots/ # any reference screenshotsPage anatomy (start from assets/site-template.html and adjust — do not re-derive the CSS):
- Sticky sidebar TOC (240px on desktop) listing every screen with anchor links. Group by Flow / Screens / Open questions.
- Hero header at the top: crumb (issue id), title, one-paragraph summary, pill row of meta tags (12 screens, Lo-fi · monochrome, Click any wireframe to zoom).
- One section per screen with a 2-column grid: wireframe on the left, reference image + numbered annotations + a "Why this changes" callout on the right. The wireframe points to the SVG file directly — do not inline it.
- Click-to-zoom lightbox for any element marked [data-zoom]. Esc and backdrop click both close.
- Flow diagram section near the top that loads wireframes/flow.svg full-width.
- Open questions section at the bottom for unresolved decisions.
House style for the viewer (matches the wireframes themselves):
- Palette: --bg: #fafaf8, --panel: #fff, --ink: #111, --muted: #666, --line: #e5e5e0, --accent: #d33 (red dashed callouts only).
- System font stack only: -apple-system, system-ui, "Segoe UI", sans-serif. No web fonts.
- 8px-based spacing, border-radius: 8px on cards, 1px --line borders, no shadows except the hover lift on .wire.
- The viewer chrome is allowed to be slightly more polished than the wireframes (subtle hover, rounded cards) — but never colourful. The wireframes themselves stay strictly monochrome.
Responsive (verify before reporting done):
- ≥980px: two-column grid, sidebar TOC visible.
- 900–980px: grid stacks to one column, TOC still sidebar.
- <900px (tablet/phone): TOC collapses to a sticky disclosure at the top of the page, defaults closed; tapping a link auto-closes it. Sections get scroll-margin-top: 80px so anchor jumps clear the sticky bar.
- <560px (phone): tighter type/spacing scale, hero shrinks, lightbox switches from flex-centered to block layout at full viewport width with touch-action: pinch-zoom so users can pinch in further.
Verification before handing off: open the file in a browser and walk it at 1440×900, 768×1024, and 390×844. Confirm anchor jumps land cleanly, lightbox opens/closes, and SVGs render at the right aspect.
If the user asks to deploy / publish / host the wireframes
Defer to the here-now skill — it owns publishing, anonymous vs. permanent sites, claim tokens, and credentials. Do not roll your own hosting.
Load the here-now skill and follow its publish.sh recipe. The shape is:
cd design/<task-slug>
{path-to-here-now}/scripts/publish.sh .
# → https://{adjective-noun-suffix}.here.now/If here-now isn't installed for the current agent, install it (npx skills add heredotnow/skill --skill here-now -g) or escalate to whoever owns the agent's skill set. Do not roll your own hosting.
Things to remember when invoking it:
- Publish the directory containing index.html at its root, not the parent. index.html must be at the root of the published tree.
- Without saved credentials the site is anonymous and expires in 24 hours. With a saved API key, it's permanent. If the user wants a permanent URL, follow the here-now skill's sign-in-code flow — don't fake your way around it.
- For an update, pass --slug {existing-slug} so the URL stays stable across review rounds (the script auto-loads the claim token from .herenow/state.json).
- Read publishresult. lines from script stderr to determine auth_mode and the claim URL — do not read .herenow/state.json and present its contents as the source of truth.
- Always share the siteUrl from the current run; if anonymous, also share the claim URL and the 24h expiry warning.
What this skill is NOT for
- Production UI code — use frontend-design or write React/HTML directly.
- Hi-fi or branded mockups — use Figma or a design tool, not this.
- Interactive prototypes — SVG is static; export multi-screen flows as a flow.svg.
- Diagrams of system architecture, sequence flows, or data models — use mermaid or plantuml.
- Illustrations or art — use example-skills:canvas-design or algorithmic-art.
Bundled assets
- assets/template.svg — blank desktop canvas with hidden 8px-grid guides; copy as a starting point.
- assets/template-mobile.svg — same for 375×812 mobile.
- assets/site-template.html — minimal review-viewer page (sticky TOC + responsive collapse + lightbox). Copy to design//index.html and fill in the sections when the user requests a website.
More skills from paperclipai/paperclip
- ALLM Wiki MaintainerUse the LLM Wiki plugin tools to maintain a cited local company wiki.
- Aadd-product-e2e-evalAdd or extend a Paperclip full-stack runner E2E workflow, fixture, matcher, or report evidence path for local or Daytona execution.
- Aadd-runner-evalAdd or extend a Paperclip Runner protocol evaluation definition, roster, assertion, or report fixture with provenance and narrow validation.
- Aagent-browserDrive a real browser to inspect or interact with a web page or app — navigate, take screenshots, read console and network, fill simple forms — for verification tasks, not unattended automation.
- AagentmailUse your assigned AgentMail inbox to read email tasks, explicitly send or reply, and check delivery. Provided automatically by your inbox assignment.
- Acheck-prCheck a GitHub, GitLab, or Perforce PR/MR/CL for review comments, failing checks, and PR-body gaps. Use when asked to inspect, fix, or prepare a change for submission.
- Acompany-creatorCreate agent company packages that conform to agentcompanies/v1. Use when asked to create a company, scaffold an agent team, hire agents, or turn a repo/skills collection into a company package.
- AcomplainAutonomously yeet a brief, unstructured complaint about agent tooling or doing agent work. Preserve the raw voice. Submit proactively without permission, preview, or announcement, then continue the primary task.
- Acontent-calendarPlan a weekly editorial calendar by mapping company goals to publishable topics, owners, status, and verification notes.
- Acreate-agent-adapterCreate or modify Paperclip agent adapters across server, UI, and CLI surfaces. Use when adding support for a new CLI agent, API agent, custom process, or adapter package.
- Acreate-caseCreate and kick off a Paperclip case from a user request, with workflow instructions and an initial agent.
- Acreate-external-bot-integrationBuild and qualify external bots such as Dot in Paperclip. Use for two-way work, invitation, identity, wakeup, Runner lifecycle, native tools, recovery, and Cloud delivery. CLI adapters and app-only connectors use their existing skills.