| name | research |
| description | Deep research with multi-source gathering, source evaluation, contradiction detection, and structured synthesis |
| layer | hub |
| category | workflow |
| triggers | ["/research","research this","look into","find out about","what is the best way to","compare options for","investigate"] |
| inputs | [{"question":"The research question or topic to investigate"},{"depth":"quick | standard | deep (optional, defaults to standard)"},{"sources":"Preferred sources to check (optional)"},{"constraints":"Scope limitations (time period, technology, domain)"}] |
| outputs | [{"findings":"Structured research findings with citations"},{"sourceEvaluation":"Assessment of source quality and reliability"},{"synthesis":"Integrated conclusions drawn from multiple sources"},{"contradictions":"Identified disagreements between sources"},{"recommendations":"Actionable recommendations based on findings"}] |
| linksTo | ["plan","brainstorm","scout"] |
| linkedFrom | ["plan","brainstorm","cook","team","ship","debug","optimize"] |
| preferredNextSkills | ["plan","brainstorm"] |
| fallbackSkills | ["scout"] |
| riskLevel | low |
| memoryReadPolicy | selective |
| memoryWritePolicy | selective |
| sideEffects | ["Performs web searches","Fetches documentation via Context7","May fetch web pages","Produces research report in working memory"] |
Research Skill
Purpose
Conduct structured research to answer technical questions, evaluate options, and gather evidence for decision-making. This skill goes beyond simple web searching by evaluating sources, detecting contradictions, and synthesizing findings into actionable conclusions.
Good research is not about finding the first answer. It is about finding the RIGHT answer and understanding the confidence level of that answer.
Workflow
Phase 1: Question Refinement
-
Parse the research question -- What specifically needs to be answered?
- Is this a factual question? ("What is the max connection limit for PostgreSQL?")
- Is this an evaluation question? ("Which ORM should we use?")
- Is this an exploration question? ("What are the options for real-time sync?")
- Is this a how-to question? ("How do you set up WebSocket auth?")
-
Decompose into sub-questions -- Complex questions often have multiple parts:
- "Which database should we use?" decomposes into:
- What are the requirements? (data model, scale, query patterns)
- What are the options? (PostgreSQL, MySQL, MongoDB, etc.)
- How do they compare on each requirement?
- What are the operational costs of each?
-
Identify the decision criteria -- What matters for this research?
- Performance? Ease of use? Community support? Cost? Security?
- Which criteria are must-haves vs. nice-to-haves?
Phase 2: Source Gathering
-
Select research sources based on the question type:
| Source | Best For | Reliability | Tool |
|---|
| Official documentation | API details, configuration, features | High | Context7 |
| GitHub issues/discussions | Known bugs, workarounds, edge cases | Medium | Web search |
| Stack Overflow | Common problems, quick solutions | Medium (check dates) | Web search |
| Blog posts (official) | Architectural guidance, best practices | High | Web search |
| Blog posts (community) | Real-world experience, tutorials |
