A Claude Code plugin that provides an AI-powered web scraping skill. It navigates a website using the Playwright MCP server, records structured interaction traces, then generates a standalone deterministic Playwright scraper in TypeScript.
/fixpoint-scraper:scrape <url> <query>
For example:
/fixpoint-scraper:scrape https://example.com/products Extract all product names and prices
The scraper operates in two phases:
Phase 1 — AI-Powered Navigation: Claude navigates the target site using the Playwright MCP browser, interacting with elements to fulfill your query. It records a structured trace of every action (clicks, fills, scrolls, etc.) along with a narrative describing the scraping strategy.
Phase 2 — Deterministic Scraper Generation: Using the recorded traces and narrative, Claude generates a standalone TypeScript Playwright script that deterministically reproduces the scraping flow. The generated scraper runs without any AI/LLM dependencies — just Playwright.
This plugin requires the Playwright MCP server to be configured in your Claude Code environment. The Playwright MCP server provides browser automation capabilities that the scraping skill depends on.
Add this plugin to your Claude Code configuration:
{
"plugins": ["github:gofixpoint/agent-plugins"]
}Apache-2.0