name: deep-research description: Conduct structured multi-source research, synthesize evidence, compare options, and produce actionable conclusions. Designed for technical, product, and market investigations. version: 1.0.0
Deep Research
Purpose
Produce high-confidence research outputs by systematically gathering, evaluating, and synthesizing evidence from multiple sources.
The goal is not just to summarize information but to:
- uncover insights
- compare alternatives
- identify tradeoffs
- generate actionable recommendations
This skill prioritizes accuracy, traceability, and structured reasoning.
Auto Trigger
Activate automatically when the user asks to:
Research
Investigate
Compare
Analyze
Study
Evaluate
Find best approach
Find alternatives
Do market research
Deep dive
Explore tradeoffs
Explain architecture
Review competitors
Understand how something works
Or when the request includes:
- pricing models
- SaaS strategy
- architecture decisions
- technology comparisons
- best practices
- documentation synthesis
- standards
- market demand
- technical feasibility
When To Use
Use this skill when the user needs:
- market research
- competitor analysis
- pricing strategy research
- architecture tradeoff analysis
- product discovery
- best practice investigation
- documentation analysis
- repo comprehension
- multi-source synthesis
Examples:
- “Research credit-based SaaS pricing models”
- “Compare Supabase vs Firebase for SaaS apps”
- “Find how modern AI tools implement SEO audits”
- “Investigate best way to implement usage billing”
When NOT To Use
Do not use this skill for:
- direct coding tasks
- UI implementation
- simple factual questions
- trivial explanations
- quick summaries
Use it when depth and rigor are required.
Research Principles
Follow these principles:
- Multi-source verification
Never rely on a single source if alternatives exist.
- Primary sources first
Prefer:
- official documentation
- engineering blogs
- academic papers
- real product examples
- Separate facts from interpretation
Clearly distinguish:
- evidence
- analysis
- conclusions
- Avoid speculation
If evidence is missing, state uncertainty.
- Prefer recent information
Technology and SaaS evolve quickly.
Research Workflow
Step 1 — Clarify the question
Define:
- the core question
- constraints
- desired outcome
- success criteria
Example:
Instead of:
“How do subscriptions work?”
Define:
“What SaaS pricing model is best for an AI audit product targeting SMEs?”
Step 2 — Define research scope
Identify:
- domains to investigate
- industries or competitors
- relevant technologies
- potential solution categories
Step 3 — Gather sources
Collect evidence from:
- official documentation
- industry reports
- engineering blogs
- product case studies
- real implementations
Look for:
- concrete examples
- real metrics
- architecture patterns
Step 4 — Extract insights
For each source:
Identify:
- key findings
- patterns
- best practices
- limitations
Step 5 — Compare approaches
Build comparison tables:
Example structure:
| Approach | Advantages | Disadvantages | Complexity | Cost |
|---|
Step 6 — Synthesize conclusions
Answer:
- what works
- what does not work
- why
- under which conditions
Step 7 — Produce recommendations
Provide:
- best approach
- alternatives
- tradeoffs
- implementation implications
Research Quality Checklist
Before finishing verify:
- multiple sources were considered
- conclusions follow from evidence
- tradeoffs are clearly explained
- recommendations are actionable
- uncertainties are disclosed
Output Structure
Always return results in this structure:
- Research Question
- Scope and Context
- Key Findings
- Evidence Summary
- Comparative Analysis
- Tradeoffs
- Recommended Approach
- Alternative Approaches
- Implementation Considerations
- Open Questions / Unknowns
Example Invocation
Explicit:
$deep-research Investigate whether SaaS credit-based billing or subscription billing is better for an AI audit platform targeting SMEs.
Implicit:
“Research pricing models for AI SaaS tools and recommend the best option.”
Research Depth Levels
If not specified assume deep analysis.
Levels:
Light
- 3–5 insights
Standard
- detailed comparison
Deep
- multiple frameworks + strategic recommendation
Special Modes
Architecture Research
Focus on:
- scalability
- reliability
- cost
- operational complexity
Market Research
Focus on:
- demand
- competitor positioning
- pricing
- customer pain points
Product Research
Focus on:
- user workflows
- UX patterns
- monetization models
Research Anti-Patterns
Avoid:
- generic summaries
- repeating documentation
- speculation without evidence
- ignoring tradeoffs
- vague recommendations