claude-deep-research-skill

Why we included this project

Anyone doing real research inside Claude Code, whether for technical evaluation or market analysis, will find this more useful than a quick chatbot summary. Give it a one-line request like "deep research on the current state of quantum computing" and it runs a pipeline that plans the work, searches multiple providers in parallel, cross-checks sources, synthesizes findings, and critiques its own draft, looping back to close gaps when evidence is thin. Four modes scale the effort from a two-minute first pass to an UltraDeep run of up to forty-five minutes, and the report lands as Markdown, styled HTML, and PDF in a dated folder, with citations written to disk so long reports survive context limits. Quality rules built in, such as minimum source counts and full bibliographies with real URLs, make the output something you can actually verify before acting on it. It is a research engine rather than a general-purpose assistant, best where citation-heavy output matters more than chat fluency.