deep-deep-research

A multi-source, adversarially-verified research pipeline that produces a cited report. It reproduces the content of Claude Code's built-in deep-research workflow (scope → search → dedup → fetch → 3-vote verify → synthesize) as an inspectable, editable skill you run with the host agent's own tools.

Executor model — who does what

You (the agent running this skill) are the executor. You supply all the LLM reasoning, the web tools, and the parallel fan-out: