Connect AI assistants to Voltaire documentation and generate custom Skills
Model Context Protocol allows AI assistants to access external tools and data sources. Use MCP servers to give Claude and other AI assistants access to Voltaire documentation and generate custom Skills for your contracts.
The primary use case for the Voltaire MCP server is generating custom Skills tailored to your specific contracts and requirements.Instead of using generic provider/contract abstractions, you can ask Claude to:
Copy
Ask AI
Using the Voltaire docs, generate a custom provider Skill that:- Includes automatic retry with exponential backoff- Caches eth_getBalance results for 12 seconds- Batches requests automatically- Is optimized for my ERC-20 token contract at 0x...
Claude will use the Voltaire documentation to generate a custom implementation based on our Skills patterns.
See Skills Philosophy to understand why Voltaire uses copyable implementations instead of rigid library abstractions.
Add the Playwright MCP server using the Claude Code CLI:
Copy
Ask AI
claude mcp add playwright npx @playwright/mcp@latest
Verify it’s connected:
Copy
Ask AI
claude mcp list
You should see playwright: npx @playwright/mcp@latest - ✓ Connected.
2
Install Playwright Browsers
Install the Chromium browser:
Copy
Ask AI
npx playwright install chromium
This downloads the browser that Playwright uses to render pages.
3
Start Local Docs Server
Start the Mintlify dev server:
Copy
Ask AI
bun run docs:dev
Server typically runs at http://localhost:3002 (or 3000 if available).
4
Use in Claude Code
In Claude Code, ask Claude to view your documentation:
Copy
Ask AI
Navigate to http://localhost:3002/primitives/bytecodeand take a screenshot
Claude can now view, screenshot, and interact with your local docs.
1
Install Playwright Browsers
Install the Chromium browser:
Copy
Ask AI
npx playwright install chromium
2
Configure Claude Desktop
Edit your Claude Desktop config file:macOS:~/Library/Application Support/Claude/claude_desktop_config.jsonWindows:%APPDATA%\Claude\claude_desktop_config.jsonLinux:~/.config/Claude/claude_desktop_config.jsonAdd the Playwright MCP server:
Edit your Amp settings file:Global (macOS):~/.config/amp/settings.jsonProject-specific:.amp/settings.json in your projectAdd the Playwright MCP server:
Ask Claude to view your documentation pages. Examples:View a specific page:
Copy
Ask AI
Can you navigate to http://localhost:3000/primitives/bytecode/analyzeand take a screenshot?
Review page structure:
Copy
Ask AI
Please visit http://localhost:3000/crypto/keccak256 and tell meif the code examples are displaying correctly.
Check navigation:
Copy
Ask AI
Navigate to http://localhost:3000 and verify the sidebar navigationincludes all primitive modules.
Validate component rendering:
Copy
Ask AI
Go to http://localhost:3000/primitives/address and check if thetabs are working for Class API vs Namespace API examples.
Claude can:
Navigate to any documentation page
Take screenshots to visualize layout
Inspect page content and DOM structure
Click elements and interact with the page
Verify links and navigation
Check responsive design at different viewport sizes
As your codebase accumulates more Voltaire examples, you’ll need the MCP server less frequently. The AI can learn from existing patterns in your code. At that point, we recommend excluding the MCP server from your context since everything the AI needs will already be in your codebase.
MCP Protocol: Model Context Protocol 1.0
Transport: stdio (standard input/output)
Browser Engine: Chromium (via Playwright)
Viewport: Default 1280x720 (configurable)
Permissions: Local network access required for localhostThe Playwright MCP server runs locally and connects to your local documentation server. No external network access required except for initial package installation.