Player not loading? Watch on YouTube
This tutorial connects Claude Code to the Firecrawl MCP server for web search and scraping. The demonstrated setup uses Firecrawl's hosted remote URL and requires no local installation. The speaker shows an API-key option and explains how to omit it to get started without a key. This is a remote web service workflow, not an offline setup.
The speaker argues that Firecrawl gives a coding assistant more useful page content than the built-in search and fetch tools, particularly on JavaScript-heavy sites. They describe clean markdown output and structured JSON extraction. To steer Claude Code toward Firecrawl, the tutorial covers explicit prompts, deny rules for the default tools, and instructions in a main or project-specific CLAUDE.md file that retain those tools as a fallback.
The examples include a Shopify page where the speaker reports that the default tools returned dollar prices and converted them incorrectly to pounds. Firecrawl's location-based scraping returns the UK prices in that demonstration. Another prompt extracts selected pricing details from Vercel's website as JSON.
The final example uses screenshots and brand assets gathered through Firecrawl to recreate the Vercel site with Astro rather than Next.js. The speaker points out missing logo shadows and mismatched text colors, while reporting that links and theme toggles work. A closing mention directs viewers to a separate Firecrawl CLI video for token-saving techniques.
