Alexey Ostrikov

It collects blog feeds and HTML indexes from AI labs and product teams over a chosen date window, filters out product announcements and model releases, and returns a digest of deep posts on agent architecture—memory, context, orchestration, evals, and long-horizon risks—with links and takeaways.

It targets engineers and researchers building autonomous AI agents who want to track fundamental publications on agent memory, long-running loops, self-improvement, and run economics from blogs like Anthropic, OpenAI, Cursor, LangChain, Amp, and Cognition without manual monitoring.

Invoke it with phrases like 'radar', 'digest', 'what's new in blogs', or 'agent research'. It uses Bash, Read, Write, WebFetch, and WebSearch permissions to gather feeds and HTML indexes for the requested period, then outputs a structured digest.

It covers blogs from Anthropic, Claude, OpenAI, Cursor, LangChain, Amp, and Cognition, keeping only deep architecture posts. It explicitly excludes corporate wikis and trackers, arXiv papers, message delivery, product announcements, and model release notes.

The skill requires an environment granting Bash, Read, Write, WebFetch, and WebSearch capabilities, since it fetches live blog feeds and HTML indexes over a user-specified date window. No other dependencies are declared in its manifest.