Ship high-fidelity wireframes faster. Build complete multi-page wireframe experiences with strategic variants, real business context, and production-ready React codeβin minutes instead of days.
Perfect for:
- Product teams testing layout hypotheses
- Designers validating UX concepts
- Agencies pitching clients with interactive prototypes
- Developers scaffolding new features
Replace static mockups with living, testable wireframes. Each variant links to business goals and user personas, making every design choice measurable.
Get started in 3 steps:
# 1. Clone and install
git clone <your-repo-url> wireframe-platform
cd wireframe-platform
npm install
# β Automatically validates setup and creates config
# 2. Create your first wireframe project
npm run init
# β Interactive wizard walks you through setup
# 3. Start developing
npm run dev
# β Opens http://localhost:8080That's it! Your wireframe platform is ready.
npm installran automatic setup validationnpm run initcreated a project workspace with your chosen template:- Blank - Minimal starting point
- Showcase - See all available components
- Example - Full project with variants and business context
npm run devstarts the dev server and shows your wireframe dashboard
After your first wireframe is running:
- Edit components in
projects/YOUR-PROJECT/src/wireframes/YOUR-PROJECT/ - Add variants to test different layouts
- Run
npm run validate:metadatato check your work - Use Chrome DevTools MCP for visual validation (pre-configured in
.mcp.json) - Deploy to Replit, Vercel, or any Node.js host
New to the platform? Check out the Quick Start Guide for a detailed walkthrough.
Setup validation fails?
- Ensure you're using Node.js v18 or higher:
node --version - Delete
node_modulesandpackage-lock.json, then runnpm installagain
npm run init doesn't work?
- Make sure you've run
npm installfirst - Check that
packages/wireframe-core/scripts/init.mjsexists
No wireframes showing up?
- Verify
metadata.jsonexists in your wireframe directory - Run
npm run validate:metadatato check for errors - Check the browser console for route conflicts
Need help? Open an issue or check CLAUDE.md for complete development guidance.
Utility commands for diagnostics and maintenance:
# Get project overview
npm run list # Shows all wireframes and variant counts
# Health check
npm run doctor # Validates Node, deps, metadata, routes, Git status
# Create new project
npm run create # Alias for npm run init (simpler)
# Validate everything
npm run validate:all # Runs all validations (metadata, routes, business context, TypeScript)
# Nuclear cleanup (with confirmation)
npm run clean:all # Cleans temp files, build artifacts, MCP cachesPro tips:
- Run
npm run doctorbefore starting work to check framework health - Run
npm run validate:allbefore committing changes - Use
npm run listto see all projects at a glance
- AGENTS.md β Operational quick-start for agents and humans (auto-generated highlights)
- Quick Start Guide β 5-minute setup walkthrough
- CLAUDE.md β Canonical development playbook (architecture, workflows, standards)
- AGENT-WORKFLOWS.md β Agent orchestration and workflow definitions
- Workflows Guide β Common development patterns
- API Reference β Complete framework API documentation
- Migration Guide β Migrate existing projects to framework
- Changelog β Framework version history
- Metadata Schema β Schema v2.0 specification
- Routing System β Dynamic routing guide
- Snapshot System β Version control for git-ignored files
- Maintenance β Temp file cleanup procedures
- DOCUMENTATION.md β Full documentation index
- Design Fundamentals β Design system principles
- Auto-generate AGENTS.md highlights:
npm run docs:build - Verify AGENTS.md is in sync:
npm run docs:check - CI enforces docs sync on PRs via
.github/workflows/docs.yml
Use slash commands in Claude Code or Codex CLI to let AI agents handle the heavy lifting. They orchestrate the right agents and CLI scripts behind the scenes, so you can stay in the chat-driven flow:
graph TB
A[Design Brief or Existing Project] --> B{Choose Path}
B -->|New Project| C[Run /start]
B -->|From URL/Screenshot| D[Run /analyze]
B -->|Iterate Existing| E[Run /start]
C --> F[Gather Business Context]
D --> G[Extract Source Insights]
E --> F
F --> H{Need Strategy Support?}
G --> H
H -->|Yes| I[Run /strategize]
H -->|No| J[Build Directly in Code]
I --> J
J --> K["Develop Components & Pages"]
K --> L[Run /validate]
L --> M{Review in Chrome DevTools MCP}
M --> U[Run /ux-review]
U -->|"Score < 80"| K
U -->|"Score >= 80"| N["Run /export (optional)"]
N --> O[Handoff or Ship]
| Command | Purpose | When to Use |
|---|---|---|
/wf |
Run full workflow end-to-end | Starting a new project from scratch |
/analyze |
Analyze brief β structured requirements | Have a brief but need strategic analysis |
/strategize |
Generate differentiated variants | Need variant hypotheses and layouts |
/validate |
Check metadata & business alignment | Before implementing or after changes |
/export |
Package artifacts (optional: Lovable) | Ready to hand off or export |
/ux-review |
Grade built variants and log UX feedback | After capturing DOM/screenshot for a variant |
/start |
Show current workflow state | Resume work or check progress |
Setup: Run npm run wf:setup -w @wireframe/core to sync slash commands with Claude Code.
Example:
/wf --project my-landing-page --brief projects/<workspace>/src/wireframes/my-landing-page/brief.txt
The workflow orchestrates agents automatically, validates alignment with business goals, and prepares everything for implementation.
# Development server
npm run dev -w projects/<project-name>
# Build for production
npm run build -w projects/<project-name>
# Validate metadata
npm run validate:metadata -w projects/<project-name>
# Lint project
npm run lint -w projects/<project-name># Build framework
npm run build -w @wireframe/core
# Lint framework
npm run lint -w @wireframe/core
# Type check framework
npm run typecheck -w @wireframe/coreClick to expand CLI reference
| Task | Command | Notes |
|---|---|---|
| Scaffold new project | npm run scaffold -w @wireframe/core -- --project <slug> [--apply-routes=root|namespace] |
Creates a workspace project with starter metadata/pages wired to the framework |
| Transcribe page β wireframe | npm run transcribe -w @wireframe/core -- --project <slug> --url=<url>npm run transcribe -w @wireframe/core -- --project <slug> --screenshot=<path> |
Generates structured section maps + metadata drafts from live pages or screenshots |
| Iterate existing wireframe | npm run iterate -w @wireframe/core -- --project <slug> --from=baseline --new=variant [--apply-routes=root] |
Clones an existing variant to jump-start a new exploration while preserving routing |
| Run agent workflow | npm run orchestrate -w projects/<project> -- --project <slug> [--prepare-prompts --force-business-context --skip-variant ...] |
Drives the end-to-end agent sequence; workspace scripts prefill a default slug, but an explicit --project keeps the command portable |
| Self-iteration loop | npm run self-iterate -w projects/<project> -- --project <slug> [--headless --isolated --auto-fix --grade-threshold <n> --non-interactive] |
Captures Chrome DevTools MCP artifacts and runs UX review + validation; prompts for config in interactive mode |
| Run UX review | npm run ux:review -w projects/<project> -- --variant <key> [--dom <path>] [--screenshot <path>] [--change-log <path>] |
Generates a ready-to-run ux-review prompt and logs JSON/Markdown outputs alongside the shared change log |
| Optional Chrome bridge | node packages/wireframe-core/scripts/mcp/devtools-http-bridge.mjs |
Launches the Puppeteer-powered MCP HTTP bridge manually (self-iteration auto-starts when needed) |
| Export business context JSON | npm run export:business-context -w @wireframe/core |
Reads Markdown context and refreshes context/temp-agent-outputs/business-context.json |
| Validate metadata schema | npm run validate:metadata -w projects/<project> |
Fails fast when metadata.json drifts from schema v2.0 |
| Migrate legacy metadata | npm run migrate:metadata -w @wireframe/core -- --project <slug> |
Converts v1 metadata and writes a .v1.backup for safety |
| Clean temp artifacts | node packages/wireframe-core/scripts/clean-context-temp.mjs [--dry-run] [--include-agent-outputs] [--include-chrome-mcp-cache] [--include-mcp-log-cache] |
Empties context/temp/ and optional MCP caches; combine flags as needed |
| Smoke-test exporter | npm run test:export-business-context -w @wireframe/core |
Ensures business-context export pipeline stays healthy |
The @wireframe/core package provides:
- WireframeHeader β Sticky navigation with mobile menu
- ErrorBoundary β React error boundary
- ProgressIndicator β Progress bar component (v2.0.1+)
- 50+ UI Primitives β shadcn-ui components (Button, Card, Input, etc.)
- Pattern Components β MetricCard, FeatureGrid, TestimonialCard, etc.
- Routing β
generateAllWireframeRoutes(),generateRoutesFromMetadata(), etc. - Metadata β
validateMetadata(),deriveVariantRoutes(),getFullRoutes(), etc. - Helpers β
cn()class merger, design tokens, etc.
- useToast β Toast notification system
- useIsMobile β Mobile viewport detection
- Vite β
createWireframeViteConfig()for project vite.config.ts - Tailwind β
createWireframeTailwindConfig()for project tailwind.config.ts - TypeScript β Base tsconfig.base.json to extend
- Metadata validation, scaffolding, orchestration, cleanup tools
Full API Reference: packages/wireframe-core/docs/API.md
- Grayscale-first β Neutral palette, minimal color
- 8px rhythm β Consistent spacing system
- 12-column grid β Max width 1280px
- WCAG 2.1 AA β Accessibility built-in
- Mobile-responsive β All breakpoints tested
See WIREFRAME-FUNDAMENTALS.md for complete guidelines.
The slash commands orchestrate these specialized agents:
- business-context-gatherer β Captures strategic goals, personas, KPIs
- brief-analyzer β Extracts requirements, sections, routing needs
- wireframe-strategist β Designs 2-3 differentiated variants
- visual-ux-advisor β Recommends layouts and interactions (optional)
- variant-differentiator β Defines hypotheses for each variant (optional)
- business-context-validator β Ensures alignment with business goals
- wireframe-validator β Grades wireframes against requirements (self-iteration)
- ux-review β Critiques built variants, logs UX findings, and gates pass/fail at 80%+
Deep dive: See AGENT-WORKFLOWS.md and templates/agents/ for complete definitions.
Why it matters: Business context transforms wireframes from arbitrary design variations into strategic experiments. Without it, you're building layout options. With it, each variant becomes a targeted hypothesis testing which personas, goals, and messaging resonate bestβwith traceability from business objectives through metadata to implementation.
How it works: The platform uses context/BUSINESS-CONTEXT.md as the strategic foundation for all wireframe decisions. Agents reference this context to ensure each variant aligns with real business objectives. Each project's metadata.json links variants to specific goalIds and personaIds, making every design decision measurable and strategically defensible.
- Initial setup: First run of
business-context-gathererscaffoldscontext/BUSINESS-CONTEXT.md(git-ignored, persistent) - Updates & revisions: Rerun gatherer with
--force-business-contextto append new intelligence - Export to structured JSON: Run
npm run export:business-contextto regenerate machine-readable format - Link to metadata: Copy goal/persona IDs into
metadata.json(businessContextandbusinessContextRef)
Result: Every wireframe variant has documented business rationale with full traceability.
- Ephemeral assets now live under
context/temp/:context/temp/screenshots/β one-off captures (Chrome DevTools MCP, reviews, etc.)context/temp/projects/<slug>/β project-specific snapshots (e.g., HTML dumps, JSON exports)
- Agent pipelines continue to write to
context/temp-agent-outputs/; leave the structure in place so orchestrator runs can resume, but feel free to clear it between sessions. - UX review transcripts live under
context/temp-agent-outputs/ux-review/<project>/(JSON + Markdown per variant) and change logs live undercontext/temp-agent-outputs/<project>/ux-review/; keep them between iterations so agents can reference prior feedback. - Use
node packages/wireframe-core/scripts/clean-context-temp.mjs --dry-runto preview cleanup, then rerun without the flag to delete. Add--include-agent-outputswhen you also want to wipe the orchestrator artifacts. - Add
--include-chrome-mcp-cacheto remove the Chrome DevTools MCP browser profile (~70β―MB+ under~/.cache/chrome-devtools-mcp). Add--include-mcp-log-cacheto purge Claude MCP logs (~/Library/Caches/claude-cli-nodejs/**/mcp-logs-chrome-devtools). - Shortcut aliases still exist in legacy projects (
npm run clean:chrome-cache,npm run clean:mcp-logs,npm run clean:all-mcp); recreate them per project if helpful. - Anything worth keeping long-term should move into
context/archive/(tracked) or the relevant project directory insidesrc/wireframes/.
- Preview with
npm run dev -w projects/<project-name> - Validate layout, interactions, and content before sign-off
npm run self-iterate -w projects/<project> -- --project <slug> --headless --isolatedAutomates validation and captures evidence:
- Auto-launches Chrome DevTools MCP bridge
- Navigates routes, collects DOM snapshots, takes screenshots
- Runs UX Review agent (grades goal alignment, accessibility, visual hierarchy, change coverage)
- Runs Wireframe Validator against captures
- If grade >= threshold AND validation passes β Done
- If
--auto-fixβ Applies fixes, iterates again - Writes artifacts to
context/temp-agent-outputs/self-iteration/<slug>/
Configuration flags:
| Flag | Purpose |
|---|---|
--grade-threshold <n> |
Override passing grade (0-100, default: 80) |
--auto-fix |
Enable automatic fix attempts |
--non-interactive |
Skip prompts (for CI/scripted use) |
--max-iterations <n> |
Override max iterations (1-10, default: 2) |
Interactive mode: When running in a terminal without --non-interactive, the script prompts for auto-fix and grade threshold preferences before starting.
Troubleshooting: packages/wireframe-core/docs/TROUBLESHOOTING-CHROME-BRIDGE.md
Located in tests/archive/ for reference only (no longer maintained).
- Framework β Vite + React 18 + TypeScript
- Styling β Tailwind CSS 3
- Components β shadcn-ui (Radix UI)
- Routing β React Router v6 (metadata-driven)
- Forms β React Hook Form + Zod
- Icons β Lucide React, @iconify/react
- Testing β Chrome DevTools MCP
- Monorepo β npm workspaces
This repository uses a monorepo structure with a reusable framework package and isolated project workspaces:
wireframe-platform/ # Root monorepo
βββ packages/
β βββ wireframe-core/ # @wireframe/core v2.0.1
β βββ src/ # Framework components, utilities, routing
β βββ scripts/ # CLI tools (init, validate, orchestrate)
β βββ configs/ # Base configs (Vite, Tailwind, TypeScript)
β βββ templates/ # Project templates (blank, showcase, example)
β βββ schemas/ # JSON schemas
β βββ docs/ # Framework documentation
β
βββ projects/
βββ YOUR-PROJECT/ # Created by `npm run init`
βββ src/wireframes/ # Your wireframe projects
β βββ YOUR-WIREFRAME/
β βββ components/
β βββ pages/
β βββ metadata.json
β βββ brief.txt
βββ vite.config.ts # Extends framework base
βββ tailwind.config.ts # Extends framework base
βββ tsconfig.json # Extends framework base
Benefits:
- β Framework evolves independently from your projects
- β Shared components available to all wireframes
- β
Update framework via
npm update @wireframe/core - β
Create new projects in < 5 minutes with
npm run init - β Zero code duplication
β Monorepo architecture with reusable framework package β Framework independence β update framework without touching projects β Multi-project support with isolated workspaces β Shared component library for rapid prototyping β Dynamic routing auto-generated from metadata β Agent workflows for brief-to-prompt generation β Chrome DevTools MCP integration for visual validation β Theme variants (dark, print, annotated) β Accessibility-first design system β Comprehensive documentation (API, Migration Guide, Workflows)
- Create feature branch:
git checkout -b feature/my-feature - Make changes and test:
npm run dev -w projects/<project-name> - Build validation:
npm run build -w projects/<project-name> - Commit with conventional format:
feat:,fix:,chore: - Create pull request with screenshots
Projects should import from framework:
// Framework components and utilities
import {
WireframeHeader,
Button,
Card,
generateAllWireframeRoutes,
useToast,
cn
} from "@wireframe/core";
// Project-specific code (use relative paths)
import { HeroSection } from "../components/HeroSection";
import metadata from "./metadata.json";Framework code uses relative imports:
// Inside packages/wireframe-core/src/
import { cn } from "../lib/utils";
import { Button } from "../ui/button";See API Reference for all available exports.
- Quick Start Guide
- AGENTS.md β Quick reference
- CLAUDE.md β Full development playbook
- API Reference β Framework API
- Workflows Guide β Common patterns
- Design Fundamentals β Design system
- Agent Workflows β Automation system
- Metadata Schema β Schema v2.0
- Changelog β Version history
- Maintenance Playbook β Operations & guardrails
- Documentation Index β All docs
v2.0.1 (2025-10-20):
- Added ProgressIndicator component
- Validated framework update workflow (zero project code changes needed)
v2.0.0 (2025-10-20):
- Extracted framework to @wireframe/core package
- Migrated to monorepo structure
- Created comprehensive documentation (API, Migration Guide, Changelog)
- Condensed CLAUDE.md (41% smaller) and AGENT-WORKFLOWS.md (71% smaller)
- Extracted 11 agent prompts to dedicated template files
See CHANGELOG.md for complete version history.
Need help? Start with AGENTS.md for quick reference, then dive into CLAUDE.md or API Reference for comprehensive guidance.