Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →You are an expert in Bolt.new by StackBlitz, the AI-powered full-stack development environment that runs entirely in the browser. You help developers go from idea to deployed app in minutes using natural language prompts — Bolt generates complete applications with frontend, backend, database, and deployment, all running in a WebContainer without local setup.
.claude/skills/terminalskills-bolt/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | 13% | 0% |
| case-19 | ✗→✓ | ▲ Improved | 2% | 0% |
| case-08 | ✓→✓ | ＝ Same ✓ | 14% | 0% |
| case-02 | ✓→✓ | ＝ Same ✓ | 36% | 0% |
| case-03 | ✓→✓ | ＝ Same ✓ | 61% | 0% |
You are an expert in Bolt.new by StackBlitz, the AI-powered full-stack development environment that runs entirely in the browser. You help developers go from idea to deployed app in minutes using natural language prompts — Bolt generates complete applications with frontend, backend, database, and deployment, all running in a WebContainer without local setup.
markdown## How Bolt Works 1. Describe your app in natural language 2. Bolt generates the full codebase (frontend + backend + database) 3. App runs live in the browser (WebContainer) 4. Iterate with follow-up prompts 5. Deploy to Netlify/Vercel with one click ## Example Prompts ### SaaS Dashboard "Build a project management dashboard with: - User authentication (email/password) - Kanban board with drag-and-drop - Team member management - Activity feed - Dark mode toggle Use React, Tailwind, and SQLite for the database." ### API + Landing Page "Create a URL shortener with: - Landing page with a form to paste long URLs - API that generates short codes - Redirect handler - Click analytics dashboard - Rate limiting (10 URLs per hour per IP) Use Astro for the frontend and Hono for the API." ### Internal Tool "Build an employee directory app: - Search by name, department, or location - Profile cards with photo, role, contact info - Department filter sidebar - CSV import for bulk adding employees - Admin panel for editing profiles"
markdown## Follow-Up Prompts (Iterate on Generated App) After initial generation: - "Add a settings page with profile photo upload" - "Make the dashboard responsive for mobile" - "Add real-time notifications using server-sent events" - "Connect to Supabase instead of SQLite" - "Add Stripe checkout for the premium plan" - "Write tests for the API endpoints" - "Fix the bug where the sidebar doesn't close on mobile" Bolt understands the full context of the generated app. Each prompt modifies the existing codebase, not starting over.
markdown## Supported Frameworks & Tools Frontend: React, Vue, Svelte, Astro, Next.js, Remix, Angular Styling: Tailwind CSS, shadcn/ui, DaisyUI, CSS Modules Backend: Node.js, Hono, Express, Fastify Database: SQLite (built-in), Supabase, Firebase, Prisma Auth: Clerk, Supabase Auth, custom JWT Deployment: Netlify, Vercel, Cloudflare Pages ## WebContainer - Full Node.js runtime in the browser - npm install works (most packages) - File system, terminal, package manager - No Docker, no local setup, no environment issues
markdown# No installation — runs in the browser # https://bolt.new # Free tier: Limited generations # Pro: $20/month (unlimited generations) # Teams: Custom pricing
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-08 | pass→pass | 11,533 | 7,557 | -34% | 1 | 1 | 0% | 1,937 | 2,211 | +14% | 0 | 0 | — |
case-01 | fail→pass | 19,529 | 20,640 | +6% | 1 | 1 | 0% | 3,526 | 3,967 | +13% | 0 | 0 | — |
case-02 | pass→pass | 14,607 | 14,905 | +2% | 1 | 1 | 0% | 2,602 | 3,532 | +36% | 0 | 0 | — |
case-03 | pass→pass | 6,148 | 5,268 | -14% | 1 | 1 | 0% | 1,094 | 1,766 | +61% | 0 | 0 | — |
case-04 | pass→pass | 14,332 | 12,297 | -14% | 1 | 1 | 0% | 2,642 | 3,207 | +21% | 0 | 0 | — |
case-05 | pass→pass | 10,251 | 7,779 | -24% | 1 | 1 | 0% | 1,674 | 2,206 | +32% | 0 | 0 | — |
case-06 | pass→pass | 7,712 | 5,794 | -25% | 1 | 1 | 0% | 1,282 | 1,803 | +41% | 0 | 0 | — |
case-07 | pass→pass | 11,469 | 10,330 | -10% | 1 | 1 | 0% | 2,181 | 2,734 | +25% | 0 | 0 | — |
case-09 | pass→pass | 3,150 | 1,671 | -47% | 1 | 1 | 0% | 547 | 1,105 | +102% | 0 | 0 | — |
case-10 | pass→pass | 11,097 | 10,795 | -3% | 1 | 1 | 0% | 2,040 | 2,854 | +40% | 0 | 0 | — |
case-11 | pass→pass | 16,048 | 12,623 | -21% | 1 | 1 | 0% | 2,671 | 3,038 | +14% | 0 | 0 | — |
case-12 | fail→fail | 11,042 | 8,812 | -20% | 1 | 1 | 0% | 2,025 | 2,468 | +22% | 0 | 0 | — |
case-13 | pass→pass | 11,172 | 5,053 | -55% | 1 | 1 | 0% | 1,813 | 1,779 | -2% | 0 | 0 | — |
case-14 | pass→pass | 10,124 | 4,412 | -56% | 1 | 1 | 0% | 1,567 | 1,703 | +9% | 0 | 0 | — |
case-15 | pass→pass | 6,930 | 5,911 | -15% | 1 | 1 | 0% | 1,104 | 1,783 | +62% | 0 | 0 | — |
case-16 | pass→pass | 11,825 | 7,832 | -34% | 1 | 1 | 0% | 1,962 | 2,281 | +16% | 0 | 0 | — |
case-17 | pass→pass | 4,144 | 2,095 | -49% | 1 | 1 | 0% | 656 | 1,198 | +83% | 0 | 0 | — |
case-18 | pass→pass | 10,710 | 10,755 | +0% | 1 | 1 | 0% | 1,856 | 3,004 | +62% | 0 | 0 | — |
case-19 | fail→pass | 13,676 | 10,454 | -24% | 1 | 1 | 0% | 2,694 | 2,759 | +2% | 0 | 0 | — |
case-20 | pass→pass | 11,789 | 9,924 | -16% | 1 | 1 | 0% | 2,333 | 3,005 | +29% | 0 | 0 | — |
case-21 | pass→pass | 11,425 | 9,390 | -18% | 1 | 1 | 0% | 2,357 | 2,751 | +17% | 0 | 0 | — |
case-22 | pass→pass | 9,885 | 9,078 | -8% | 1 | 1 | 0% | 1,887 | 2,514 | +33% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 22 cases were attempted. The headline lift of +9 percentage points is the difference between those two pass rates over the 22 comparable cases.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.