Conduct multi-source research with parallel subagents and adversarial validation.
npx skills add https://github.com/kirillgreen/skills --skill deep-research-kirillgreen
Please help me install this Agent Skill. Skill: deep-research Source: https://github.com/kirillgreen/skills/tree/main/deep-research Command: npx skills add https://github.com/kirillgreen/skills --skill deep-research-kirillgreen
This skill solves the problem of inconsistent, shallow AI research by enforcing a structured, multi-step process that ensures every factual claim is verified, cited, and cross-referenced.
Invoke the deep-research skill by typing /deep-research followed by your research topic to begin a thorough investigation.
High-intent search queries and answers about installing and using this skill.
Multi-source research with adversarial validation uses parallel subagents to gather information from 10-20+ sources, then employs a contrarian agent to challenge conclusions and identify blind spots. This ensures findings survive expert scrutiny.
Source credibility assessment grades sources on authority and independence to prevent marketing fluff or biased content from being presented as fact. This filtering is applied automatically during the evidence-gathering phase.
Automated research suits high-stakes competitive analysis, new technology evaluation, and complex regulatory investigation. It enforces a structured, multi-step process ensuring every factual claim is verified, cited, and cross-referenced.
Inline citations and independent evidence chains are generated by cross-referencing multiple parallel sources during the research process. This satisfies accuracy requirements for high-stakes decision-making scenarios.
AI research returns shallow results without a structured process because it lacks adversarial review and source credibility filtering. Enforcing multi-step verification with a contrarian agent ensures every factual claim is cross-referenced.
Multi-agent research is not suited for simple, single-answer queries or situations where speed is prioritized over rigorous source verification and adversarial validation. It is designed for complex, high-stakes investigations requiring depth.
SoufianouassifResearch complex topics across five focus areas and produce cited reports.
zachmayerOrchestrate parallel multi-model reviews across Codex, Gemini, and two Claude configurations.
luisgustavooliveiraExtract text and structured data from web pages using Python libraries.
caoronglinIntegrate multi-source web and academic searches with structured summarization.
