deep-research_skill

Overview

This skill conducts exhaustive, citation-rich research on any topic using web search, browser automation, documentation APIs, and codebase exploration. It prioritizes accuracy over speed, cross-references claims across independent sources, and outputs structured findings with confidence levels and source-quality assessments. Use it when you need a deep dive, fact-checking, or a comparative evaluation based on verifiable evidence.

How this skill works

The skill begins by scoping the core question, required depth, and recency constraints. It discovers sources via multiple query phrasings, fetches full content (including interactive pages when needed), extracts claims and metadata, and cross-verifies each major fact against independent sources. Results are synthesized into a clear summary with numbered citations, confidence levels, conflict analysis, and a source-quality table. Uncertainties and gaps are flagged explicitly rather than hidden.

When to use it

• Ask for a deep dive: "research X" or "deep dive into Y"
• Fact-checking contentious or unclear claims
• Comparing tools, vendors, or technical approaches
• Investigating current state or recent changes (APIs, regulations, news)
• Exploring implementation details inside a codebase or documentation

Best practices

• Define the core question and recency needs before research starts
• Require at least two independent sources for major claims
• Prefer primary, authoritative sources and note their version/date
• Report conflicts with citations and explain why they may differ
• Provide explicit confidence levels and document remaining gaps

Example use cases

• Produce a citation-backed summary of regulatory changes affecting a technology
• Compare three SaaS vendors with pros/cons, evidence, and confidence scores
• Verify claim provenance and contradictions in a viral article
• Investigate an API behavior across versions using docs and code examples
• Compile a literature-backed overview of a scientific or technical topic