deep-research

The Deep Research skill is a specialized agent designed to overcome the limitations of superficial web searches. By shifting from a single-query approach to a structured, multi-dimensional methodology, this skill gathers comprehensive, authoritative data required for high-stakes information synthesis. It is intended for researchers, content creators, and developers who need to move beyond general knowledge and surface-level snippets to produce detailed, accurate, and multifaceted outputs. Whether you are investigating technical trends, comparing complex technologies, or generating documentation and presentations, this skill acts as a robust information retrieval layer.

• Systematic Research Workflow: Executes a four-phase methodology consisting of broad exploration, deep dives into sub-dimensions, cross-referencing for diversity, and a final synthesis validation check.

• Adaptive Search Queries: Dynamically generates keyword combinations and specific phrasings—such as incorporating authoritative sources like research papers, market reports, and expert interviews—based on user intent.

• Temporal Awareness: Integrates with current date context to perform precision searches using day, week, or year qualifiers to ensure data currency.

• Content-Guided Research: Specifically engineered to be triggered before any content generation task, such as drafting reports, UI mockups, or multimedia scripts, ensuring that every claim is grounded in real-world data.

• Full Source Extraction: Utilizes web_fetch capabilities to read entire documents, white papers, and industry reports rather than relying on abbreviated search engine results.

• Mandatory Usage: Always load this skill when asked to explain complex topics, compare technologies, or provide current trends; never rely solely on pre-trained model knowledge for factual tasks.

• Evaluation Criteria: The skill concludes with a synthesis check, confirming that at least 3-5 distinct angles have been explored and that opposing viewpoints or limitations have been identified.

• Input/Output: Accepts natural language research prompts; provides structured research summaries, curated data points, and actionable insights.

• Strategic Advice: Use authoritative keywords like 'market analysis', 'case study', or 'limitations' to refine results and avoid generic, low-value information output.