Research Modes

Research Modes Skill

Version: 1.1 Created: 2026-02-02 Author: Manus Purpose: Structured approaches for deep and wide research tasks

Overview

This skill encodes two complementary research modes: Deep Research (focused, comprehensive investigation of a specific topic) and Wide Research (broad scan across multiple topics to identify patterns and opportunities). Use this skill to conduct efficient, high-quality research that produces actionable insights.

Output templates for all phases are in references/research-output-templates.md.

When to Use This Skill

• Planning a new feature or system architecture
• Investigating a technical problem or design challenge
• Exploring competitive landscape or market trends
• Synthesizing learnings from multiple sources
• Making informed decisions based on evidence

Research Mode Selection

Deep Research Mode

Use when:

• You need comprehensive understanding of a specific topic
• The decision depends on technical details
• You're evaluating a complex system or architecture
• You need to become an "expert" in a narrow domain

Characteristics:

• Focused scope (1-3 related topics)
• Multiple sources per topic (5-10+)
• Deep analysis and synthesis
• Produces Research Brief or Research Synthesis document

Timeline: 2-8 hours

Wide Research Mode

Use when:

• You're exploring a new problem space
• You need to identify patterns across multiple domains
• You're scouting for inspiration or best practices
• You want to understand the landscape before diving deep

Characteristics:

• Broad scope (10-50+ topics)
• Few sources per topic (1-3)
• Pattern recognition and clustering
• Produces Landscape Map or Opportunity Matrix

Timeline: 1-4 hours

Deep Research Mode

Phase 1: Define Scope (15-30 minutes)

Questions to answer:

1. What is the core question I'm trying to answer?
2. What decision will this research inform?
3. What level of detail do I need?
4. What are the boundaries (in scope vs. out of scope)?
5. What success criteria will I use?

Output: Research Brief — see references/research-output-templates.md

Phase 2: Source Discovery (30-60 minutes)

Methods:

• Search for academic papers, technical documentation, blog posts
• Identify authoritative sources (official docs, research labs, industry leaders)
• Look for case studies, implementations, and real-world examples
• Check GitHub repositories, open-source projects, and code examples

Quality Filters:

• Recency (prefer sources from last 2-3 years unless historical context is needed)
• Authority (prefer official docs, peer-reviewed papers, recognized experts)
• Relevance (directly addresses the research question)
• Depth (provides technical details, not just overviews)

Output: Source List (10-20 sources) — see references/research-output-templates.md

Phase 3: Deep Reading & Note-Taking (1-3 hours)

Process:

1. Read each source with the research question in mind
2. Extract key insights, claims, and evidence
3. Note disagreements or contradictions between sources
4. Identify patterns and themes
5. Flag open questions or gaps

Note-taking structure per source:

• Main argument (1-2 sentences)
• Key insights (bulleted)
• Evidence (data points, studies, examples)
• Disagreements with other sources
• Open questions raised
• Exact quotes with page or section reference
• Relevance to research question

Phase 4: Synthesis & Analysis (1-2 hours)

Questions to answer:

1. What are the major themes or patterns?
2. What do most sources agree on?
3. Where do sources disagree, and why?
4. What are the tradeoffs or tensions?
5. What gaps remain in the knowledge?

Output: Research Synthesis document — see references/research-output-templates.md

Phase 5: Validation (30-60 minutes)

Questions to ask:

1. Did I answer the research question?
2. Are my recommendations actionable?
3. Did I consider counterarguments?
4. Are there gaps in my reasoning?
5. Would someone else reach the same conclusion?

Validation Methods:

• Review against success criteria
• Check for confirmation bias (did I only seek supporting evidence?)
• Test recommendations against edge cases
• Share with a peer for feedback (if available)

Wide Research Mode

Phase 1: Define Landscape (15-30 minutes)

Questions to answer:

1. What problem space am I exploring?
2. What are the boundaries of this landscape?
3. What am I looking for (patterns, tools, approaches)?
4. How will I know when I've covered enough ground?

Output: Landscape Brief — see references/research-output-templates.md

Phase 2: Rapid Scanning (1-2 hours)

Process:

1. Search broadly across the problem space
2. Skim sources quickly (5-10 minutes per source)
3. Extract 1-3 key insights per source
4. Tag sources by category, theme, or approach
5. Move on quickly (don't get stuck in details)

Output: Tagged source list (20-50 sources) organized by category

Phase 3: Pattern Recognition (30-60 minutes)

Questions to answer:

1. What categories or clusters emerge?
2. What approaches are most common?
3. What innovations or outliers stand out?
4. What gaps or opportunities exist?

Output: Landscape Map — see references/research-output-templates.md

Phase 4: Opportunity Matrix (30-60 minutes)

Process:

1. Identify potential approaches or solutions
2. Evaluate each on key dimensions (effort, impact, risk, novelty)
3. Plot on a 2x2 matrix (e.g., effort vs. impact)
4. Prioritize based on goals

Output: Opportunity Matrix — see references/research-output-templates.md

Hybrid Research Mode

Use when:

• You need both breadth and depth
• The problem space is large and complex
• You're making a high-stakes decision

Process:

1. Start with Wide Research (2-4 hours)
2. Identify 2-3 promising areas
3. Conduct Deep Research on each area (2-4 hours per area)
4. Synthesize findings across all areas
5. Make recommendation

Timeline: 1-2 days

Best Practices

• Pick the mode before you start: Switching from Wide to Deep mid-session wastes effort — the source collection strategies and note-taking structures are different. Spend 5 minutes on Phase 1 scope definition before touching any source.
• Set a time limit per phase: Without time-boxing, Source Discovery and Deep Reading expand indefinitely. A Research Brief produced at a deadline is more useful than a perfect synthesis that never ships.
• Record disagreements, not just agreements: Sources that contradict each other are the most informative signal. If all your sources agree, you may have sampled from a single school of thought.
• Separate collection from synthesis: Do not write conclusions while reading sources. Complete note-taking first, then move to synthesis as a distinct phase. Blending them produces summaries masquerading as analysis.
• Validate against the original question: Before finalizing, re-read the research question from Phase 1 and confirm the synthesis actually answers it. Scope creep commonly produces thorough research on the wrong question.
• Label confidence levels: Distinguish between well-supported findings (multiple independent sources), moderately supported (single authoritative source), and inferred (your interpretation of ambiguous evidence). Unlabeled research papers overconfidence to downstream consumers.

Research Quality Checklist

Before finalizing research, verify:

Scope & Focus

• Research question is clearly defined
• Scope boundaries are explicit
• Success criteria are measurable

Source Quality

• Sources are authoritative and recent
• Multiple perspectives are represented
• Contradictions are acknowledged
• Bias is considered

Analysis Depth

• Key findings are supported by evidence
• Tradeoffs and tensions are identified
• Open questions are flagged
• Recommendations are actionable

Synthesis

• Patterns are clearly articulated
• Insights are connected to the research question
• Gaps and opportunities are identified
• Next steps are defined

Common Pitfalls to Avoid

• Scope Creep: Starting focused, ending scattered — define boundaries upfront
• Confirmation Bias: Only seeking supporting evidence — actively seek counterarguments
• Analysis Paralysis: Reading forever, never synthesizing — set time limits
• Surface Skimming: Reading titles, not content — take structured notes
• No Synthesis: Collecting info, not building understanding — answer the research question

Usage Instructions

1. Read this skill before starting research
2. Choose the right mode (Deep, Wide, or Hybrid)
3. Define scope clearly (research brief or landscape brief)
4. Follow the phase structure for your chosen mode
5. Take structured notes using the templates in references/research-output-templates.md
6. Synthesize findings into actionable insights
7. Validate against success criteria

Output

• Deep mode: Research Brief (scope + question) + Research Synthesis document saved to the project's docs/research/ or scouts/ directory
• Wide mode: Landscape Map + Opportunity Matrix document saved to the same location
• Format: Markdown file named [date]_[topic]_research.md or [date]_[topic]_landscape.md

Examples

Scenario 1: "Research AI agent memory compression techniques deeply" → Research Brief defining question + Research Synthesis with findings across 8-12 sources, confidence ratings, tradeoffs table, and ranked recommendations

Scenario 2: "Map the landscape of open-source design system tools" → Landscape Brief + Landscape Map categorizing 30+ tools by personality/density/use-case + Opportunity Matrix scoring the top options

Edge Cases

• Topic too broad for deep mode: Split into 2-3 focused sub-questions, run Deep Research on each, then synthesize
• Fewer than 3 credible sources found: Note the evidence gap explicitly in the synthesis; do not inflate confidence
• Conflicting authoritative sources: Surface the contradiction as a finding, do not resolve it artificially
• Time-boxed request (e.g., "quick scan"): Default to Wide Research with Phase 2 limited to 10-15 sources; label output as a preliminary scan

Anti-Patterns

• Running Wide Research when a specific decision needs Deep Research — the outputs are structurally different and not interchangeable
• Starting Phase 4 synthesis before completing source reading — partial data produces overconfident recommendations
• Reusing a prior Research Brief for a different question without redefining scope — old scope constraints silently distort new findings
• Writing the synthesis as a summary of each source in order — synthesis must be organized by themes, not by source

Related Skills:

• specification-writer — Research informs specifications
• seed-extraction — Extract seeds from research findings
• memory-garden — Document research in memory for future reference

Last Updated: 2026-04-08 Maintained By: Manus Status: Active