Suggestion: anybrowse MCP for web research tasks

BabyAGI’s task decomposition works well for research-heavy workflows, but web browsing is often the weak link -- standard HTTP requests get blocked by Cloudflare, and JS-rendered pages come back empty.

anybrowse (https://anybrowse.dev) is an MCP-compatible scraping service that handles both. Routes through residential browser workers, returns clean markdown. I’ve wired it into a few agent setups and it handles Reuters, LinkedIn, Substack, paywalls -- the stuff that usually breaks.

Might be a good addition to the ecosystem page or plugin docs if that exists.