Web2cli – Every website is a Unix command
Agents scrape websites 100x faster than Chromium, no API keys or auth nightmares.
Production-oriented authenticated web access service with hosted link flows, signed launch tokens, scoped extraction, and agent-facing APIs.
Plaid comparison oversells it — this is Browserless with CSS selector blueprints.
Developers building AI agents that need authenticated web data
Browse AI · Apify · ScrapingBee
Plaidify is open-source infrastructure that turns any login-protected website into a REST API. You drop a JSON "blueprint" into the connectors folder — CSS selectors for username, password, submit, and what to extract — and you get a structured JSON endpoint. No scraper code to write.
It's designed specifically for the AI agent era. The repo includes ready-to-copy integration code for LangChain, CrewAI, OpenAI function calling, and a planned MCP server so Claude/GPT can use it as a native tool.
What's built today: FastAPI with 19 endpoints, JWT auth, Fernet credential encryption at rest, Alembic migrations, 53 tests at 80% coverage, Docker support, CI pipeline. What's NOT built yet: the Playwright browser engine (it returns simulated responses today). We're building in public.
The hardest and most impactful open contribution is the Playwright engine — replacing the stub with real browser automation. Blueprint contributions (write a JSON file for a public test site) are easy first issues.
MIT licensed. Python 3.9+. Self-hosted — credentials never leave your infrastructure.
Agents scrape websites 100x faster than Chromium, no API keys or auth nightmares.
Local Rainforest alternative using Tauri desktop app and Playwright MCP.
Pattern guides for Playwright tests, but this is context-injection—a known workaround, not a platform shift.
Budget gates for CI failures beat Lighthouse's pass/fail binary.
Self-healing Playwright tests via AI agent repair, but Playwright Inspector and Cypress already auto-fix selectors.
Landing page errors on load—cannot evaluate product beyond the concept.