deep-research

To turn vague questions into structured web research, this Skill converts open-ended prompts into multi-query searches, fetches primary sources, and outputs numbered citations with relevance notes. It applies confidence scoring to each finding so you get evidence-based answers.

Confidence scoring in comparative analysis research rates each finding as HIGH, MEDIUM, or LOW based on source quality and relevance. This scoring mechanism helps you quickly evaluate the trustworthiness of evidence before making decisions.

To validate best practices for unfamiliar technologies, request a deep research investigation with depth selection. The Skill performs multi-query web searches, fetches primary sources for deep reads, and provides confidence-rated findings with source citations.

You can reuse cached web research findings because the Skill checks prior results saved in the agent memory directory. It reuses cached findings within 30 days to avoid redundant searches and accelerate the investigation process.

The research tool integrates findings by updating the local knowledge markdown file, logging research activities, and optionally syncing structured findings into a Graphify project when a graph exists. This ensures your knowledge base stays current.

The best way to control research depth is using the Skill's depth selection feature, which chooses quick, standard, or deep investigation modes based on your request. Deep reads trigger optional primary-source fetching for more thorough validation.