Deep Research Executor Suite

Comprehensive research execution framework that conducts multi-source deep dives on technical topics, synthesizes findings, evaluates source credibility, and produces structured research reports.

When to Use This Skill

Choose Deep Research Executor when:

• Evaluating technologies, libraries, or frameworks for adoption
• Investigating complex bugs that require understanding underlying systems
• Preparing technical documentation requiring authoritative sources
• Comparing architectural approaches with evidence-based analysis
• Building knowledge bases on emerging technologies

Consider alternatives when:

• The answer is in the project's existing documentation
• A quick web search would suffice for simple factual lookups
• You need real-time monitoring rather than point-in-time research

Quick Start

Copy
# Start a deep research session
claude skill activate expert-deep-research-executor-suite

# Research a technology
claude "Deep research: Compare WebSocket vs SSE vs Long Polling for real-time features"

# Investigate a domain
claude "Research the current state of WebAssembly in server-side applications"

Example Research Report

Copy
## Research Report: Edge Computing for API Latency Reduction

### Executive Summary
Edge computing reduces API latency by 40-60% for geographically distributed users
by processing requests at CDN edge nodes rather than centralized origin servers.

### Key Findings
1. **Cloudflare Workers**: V8 isolate-based, <1ms cold start, 30s CPU limit
   - Source: Cloudflare documentation, benchmark studies
   - Confidence: High (well-documented, widely adopted)

2. **Deno Deploy**: Built on Deno runtime, TypeScript-native, global deployment
   - Source: Deno documentation, community benchmarks
   - Confidence: Medium (newer platform, smaller adoption)

3. **Vercel Edge Functions**: Next.js integrated, streaming support
   - Source: Vercel documentation, production case studies
   - Confidence: High (production-proven at scale)

### Comparison Matrix
| Feature | CF Workers | Deno Deploy | Vercel Edge |
|---------|-----------|-------------|-------------|
| Cold Start | <1ms | ~5ms | ~3ms |
| Max Duration | 30s | 50ms (free) | 30s |
| Runtime | V8 | Deno | V8 |
| Pricing | $0.50/M req | $0.50/M req | Included |

### Recommendations
For most web applications: Cloudflare Workers (maturity, ecosystem).
For Next.js projects: Vercel Edge Functions (integration).
For Deno/TypeScript: Deno Deploy (native support).

Core Concepts

Research Methodology

Phase	Activities	Deliverable
Scoping	Define research questions, boundaries, success criteria	Research brief
Source Identification	Find primary docs, papers, benchmarks, case studies	Source registry
Data Collection	Extract relevant information from each source	Raw findings
Analysis	Cross-reference findings, identify patterns and conflicts	Analysis notes
Synthesis	Combine findings into coherent narrative with evidence	Draft report
Validation	Verify claims, check for bias, assess confidence levels	Final report

Source Credibility Framework

Tier	Source Type	Confidence Weight
Tier 1	Official documentation, peer-reviewed papers	1.0
Tier 2	Production case studies, official benchmarks	0.8
Tier 3	Reputable tech blogs, conference talks	0.6
Tier 4	Community posts, forum discussions	0.4
Tier 5	Unverified claims, marketing materials	0.2

Copy
interface ResearchReport {
  topic: string;
  questions: string[];
  sources: Source[];
  findings: Finding[];
  synthesis: string;
  recommendations: Recommendation[];
  confidenceScore: number; // 0-1 overall confidence
  limitations: string[];
}

interface Finding {
  claim: string;
  evidence: string;
  source: Source;
  confidence: 'high' | 'medium' | 'low';
  conflictsWith?: Finding[];
}

Configuration

Parameter	Description	Default
depth	Research depth: shallow, moderate, deep, exhaustive	deep
max_sources	Maximum number of sources to consult	20
min_confidence	Minimum confidence threshold for including findings	0.4
include_counterarguments	Include opposing viewpoints and limitations	true
output_format	Report format: markdown, pdf, json	markdown
recency_bias	Prefer recent sources (weight by publication date)	true

Best Practices

1. Start with primary sources, then expand outward — Official documentation and original research papers provide the most accurate foundation. Build on these with secondary analysis and community experience rather than starting with blog posts and opinions.

2. Track conflicting information explicitly — When sources disagree, document both positions with their evidence. Don't silently pick one perspective. Conflicts often reveal nuance, edge cases, or different context assumptions.

3. Assess recency alongside credibility — A highly credible source from 3 years ago may be outdated for fast-moving technologies. Note publication dates and check for updates or corrections to original findings.

4. Separate facts from interpretations in your notes — Mark direct quotes and verifiable data points differently from analysis and opinions. This prevents inadvertently presenting someone's interpretation as established fact.

5. Define "good enough" criteria upfront — Deep research can expand indefinitely. Set clear criteria for when you have sufficient evidence to answer your research questions and stop collecting when those criteria are met.

Common Issues

Research scope creeps as interesting tangents emerge. Maintain a strict research brief with 3-5 specific questions. When tangential topics arise, note them in a "future research" section rather than following them immediately. Time-box each research phase and enforce transitions.

Conflicting sources make it impossible to reach firm conclusions. Resolve conflicts by examining the methodology behind each claim. Check sample sizes, testing conditions, and potential biases. When resolution is impossible, present the range of findings with confidence intervals rather than forcing a single answer.

Research findings become outdated before the report is complete. For fast-moving topics, include version numbers, dates, and caveats about expected changes. Structure reports so individual sections can be updated independently. Link to live sources rather than only capturing snapshots.