in-depth-research

name	in-depth-research
version	1.0.0
author	G-HunterAi
license	MIT
description	Conduct exhaustive multi-source investigation with methodology tracking, source evaluation, and iterative depth.
tags	["research","analysis","investigation","methodology","sources","synthesis"]
platforms	["all"]
category	research
emoji	🔬

In-Depth Research

Conduct exhaustive multi-source investigation with systematic methodology, source evaluation, and iterative depth. Move beyond surface-level searches to thorough exploration that answers complex questions with documented evidence.

When to Use

Load this skill when:

• Complex questions requiring investigation across multiple sources
• Decision-critical research: The answer will drive important decisions
• Topic unfamiliar to you: Need to build deep understanding from scratch
• Contradictions exist: Sources disagree and you need to reconcile them
• Credibility matters: Must distinguish reliable sources from unreliable ones
• Methodology important: Need to document your process and reasoning
• Trend research: Understanding evolution of a topic over time
• Market/competitive analysis: Evaluating options systematically

Examples:

• "Should we adopt this technology? Research pros, cons, adoption rates, costs."
• "What's the current state of X field? Who are key players? What's the trajectory?"
• "Is this claim true? What does reliable research say?"
• "How do these approaches compare? What are tradeoffs?"

When NOT to Use

Do NOT use this skill for:

• Quick facts: Simple factual questions (→ just search)
• Synthesizing existing docs: Combining internal documents (→ use Synthesize)
• Ongoing monitoring: Continuous tracking of changes (→ use Digest/automation)
• Surface-level answers: "Here's what Wikipedia says" (→ standard search sufficient)

Prerequisites

• Web search access (Google, Bing, or similar)
• Web fetch capability (ability to retrieve full articles)
• Time budget (research takes 30 min to hours depending on depth)

Protocol

Scope → Search → Evaluate → Deepen → Synthesize → Document → Deliver

1. Scope

Before starting any search, clarify:

• What exactly is the question? (Reframe vague questions into specific, answerable queries)
• What depth is needed? (Quick/Standard/Thorough/Exhaustive)
• What decision does this enable? (Understand the stakes)
• Time/effort budget? (Realistic time allocation)

2. Search

Multi-vector search approach:

• Start broad: General search engines, Wikipedia, overview articles to map territory
• Go specific: Domain-specific databases, academic sources, industry reports
• Follow trails: Citation checking, author searches, reverse searches for who cites papers
• Alternative vectors: Forums (Reddit, HN), expert blogs, social media threads, news archives

Track everything: Every source consulted, every query used. Nothing unattributed.

3. Evaluate

For each significant source, assess:

Factor	Check
Authority	Who wrote it? Credentials? Institutional backing?
Currency	When published? Still accurate?
Evidence	Claims supported by data? Methodology shown?
Purpose	Inform? Persuade? Sell? Academic?
Bias	Funding sources? Conflicts of interest?
Corroboration	Do independent sources confirm?

Flag low-credibility sources. Weight findings accordingly.

4. Deepen

Research is iterative. Initial findings reveal new questions:

• Follow promising threads that emerge
• Fill identified gaps
• Chase contradictions to source
• Stop when: answer is clear, returns diminish, or budget exhausted

Document decision to stop and why.

5. Synthesize

Merge findings:

• Reconcile contradictions explicitly
• Weight findings by source quality
• Note confidence levels
• Identify remaining unknowns
• Highlight areas of consensus vs. disagreement

6. Document

Research trail matters for credibility:

• Sources consulted (with links, dates accessed)
• Search queries and strategies used
• Why certain sources were weighted higher
• What was NOT found (gaps and limitations)
• When research was stopped and why

7. Deliver

Format per user needs (see references/output-formats.md):

• Executive: BLUF + key findings + confidence
• Standard: Full findings with clear sourcing
• Academic: Complete methodology + full citations
• Working doc: All findings for further iteration

Output Format (Default)

🔬 DEEP RESEARCH: [Topic]

⚡ ANSWER
[Direct answer to the question — 2-3 sentences]

📊 CONFIDENCE: [High/Medium/Low]
[Why this confidence level]

🔍 KEY FINDINGS
• [Finding 1] — [source credibility: Tier X]
• [Finding 2] — [source credibility: Tier X]
• [Finding 3] — [source credibility: Tier X]

⚠️ CAVEATS
• [Important limitation or uncertainty]

🕳️ GAPS
• [What couldn't be determined]

📚 SOURCES ([count])
[Numbered list with credibility notes]

🔎 METHODOLOGY
[Brief: what was searched, how sources were evaluated]

Depth Levels

Choose depth based on stakes:

Level	Time	Sources	When
Quick	5-10 min	3-5	Simple factual questions
Standard	30-60 min	8-15	Most research requests
Thorough	2-4 hours	20-30	Important decisions
Exhaustive	Days	50+	Critical, high-stakes decisions

Confirm depth with user before starting. Adjust if findings warrant deeper investigation.

Example Research Prompts

Quick Research

"What are the main benefits and drawbacks of using TypeScript in a startup?"

• Search: StackOverflow, dev blogs, GitHub discussions
• Time: 20-30 minutes
• Sources: 5-8

Standard Research

"What's the current landscape of AI coding assistants? Compare capabilities, pricing, and adoption."

• Search: Industry reports, benchmarks, user reviews, pricing pages
• Time: 45-90 minutes
• Sources: 10-20

Thorough Research

"Should we migrate from React to Vue? Systematic comparison of community, ecosystem, hiring difficulty, performance."

• Search: Framework docs, benchmarks, industry surveys, job market, GitHub metrics
• Time: 3-4 hours
• Sources: 25-40

Source Hierarchy Reference

See references/sources.md for detailed source evaluation framework.

Quick reference:

• Tier 1 (highest): Primary research, official data, peer-reviewed studies
• Tier 2 (high): Expert analysis, systematic reviews, quality journalism
• Tier 3 (medium): Secondary sources, educational content, trade journals
• Tier 4 (lower): General news, Wikipedia, aggregators
• Tier 5 (lowest): Forums, blogs, user content (useful for practitioner knowledge)

Search Strategy Reference

See references/methodology.md for detailed techniques:

• Boolean operators and advanced search
• Domain-specific databases
• Citation following (forward and backward)
• Author tracking
• Forum mining (Reddit, HN, forums)
• Detecting when to stop research

Works Well With

• web-scraper: Extract data from websites you discover
• data-analysis: Analyze research data and statistics
• news-aggregator-skill: Find and track relevant news sources
• workflow-orchestrator: Schedule recurring research on topics

Common Research Scenarios

Evaluating a Technology

1. Search for official documentation and benchmarks
2. Find independent benchmarks and comparisons
3. Check adoption rates (GitHub stars, StackOverflow mentions, job postings)
4. Read case studies from companies using it
5. Find user reviews and community sentiment
6. Evaluate documentation quality and learning curve

Market Research

1. Find industry reports (Gartner, McKinsey, analyst firms)
2. Check government statistics for market size
3. Identify major players (SEC filings, company websites)
4. Research pricing and business models
5. Look for growth trends (news, analyst predictions)
6. Find customer reviews and satisfaction metrics

Academic/Historical Research

1. Use Google Scholar for peer-reviewed papers
2. Check citation counts (who cited this work?)
3. Find primary sources that papers reference
4. Look for systematic reviews or meta-analyses
5. Check for subsequent research that builds on or contradicts findings
6. Find author's other work for context

Competitive Analysis

1. Gather competitor info (public records, websites, pricing)
2. Find user reviews and comparisons
3. Check job postings (hiring strategy, locations)
4. Review funding/investment news
5. Analyze their content and positioning
6. Find third-party analyst reports

Confidence Assessment

How to determine confidence level:

• High: Multiple independent Tier 1-2 sources agree, no significant contradictions
• Medium: Mix of Tier 2-3 sources with some agreement, minor contradictions explained
• Low: Limited sources (Tier 3-5), significant disagreement, key unknowns remain
• Cannot determine: Insufficient sources or contradictory findings without resolution

Always state WHY you assessed confidence at a particular level.

Error Handling

Issue	Cause	Solution
Paywalled source blocks access	Premium content behind subscription	Note as "paywall-blocked", find alternative sources, cite what's accessible
Source returns 404/timeout	Dead link or server down	Use Wayback Machine or alternative archives; mark source as unreachable with date
Contradictory sources	Two credible sources disagree	Present both positions, note the disagreement, assess which has stronger evidence
No sources found for claim	Obscure or novel topic	Explicitly state "no corroborating sources found" with confidence: Low
Source language barrier	Key source in non-English language	Use translation tools, note "machine-translated" in citation

Red Flags to Investigate

Be suspicious when:

• Only one source makes a claim (find corroborating sources)
• Source has obvious financial interest (note bias)
• No methodology or data shown (question credibility)
• Extraordinary claims, ordinary evidence (unlikely)
• Consensus claimed but not demonstrated (verify claim)
• Source quotes another source without checking original (follow chains)

---

References: see references/ directory for methodology, sources, and output formats