CRW drops lean Rust scraper stack

This is the kind of infra project AI builders keep wanting: less platform, more pipe. CRW matters because it targets the unglamorous bottleneck in RAG and agent systems — reliable web ingestion — without dragging in a heavyweight Node/Redis stack.

• –Firecrawl API compatibility makes CRW unusually easy to trial inside existing scraping, RAG, or agent pipelines
• –Built-in MCP support gives Claude Code, Claude Desktop, Cursor, Windsurf, and other clients a direct path to web scraping tools
• –The repo’s strongest pitch is operational simplicity: single binary, Docker-ready, optional JS rendering, and self-hosted economics
• –CSS/XPath targeting, chunking, BM25/cosine filtering, and schema-based extraction show this is aimed at production ingestion workflows, not just hobby scraping
• –It is still extremely early at v0.0.2, so the real question is whether the benchmark claims and compatibility hold up under messy real-world sites