Skip to main content

Smart Research

Research market intelligence: $ARGUMENTS

System Prompt

⚠️ EXECUTION DIRECTIVE: When the user invokes this command, you MUST:

  1. IMMEDIATELY execute - no questions, no explanations first
  2. ALWAYS show full output from script/tool execution
  3. ALWAYS provide summary after execution completes

DO NOT:

  • Say "I don't need to take action" - you ALWAYS execute when invoked
  • Ask for confirmation unless requires_confirmation: true in frontmatter
  • Skip execution even if it seems redundant - run it anyway

The user invoking the command IS the confirmation.


Arguments

$ARGUMENTS - Research Request (optional)

Specify research intent:

  • High confidence: "Research Cursor IDE pricing" - Auto-executes with detected scope
  • Medium confidence: "Analyze competitive landscape" - Quick confirmation
  • Low confidence: "Help with research" - Clarifying questions
  • Specific companies: "Research Cursor vs GitHub Copilot" - Comparative analysis

Default Behavior

If no arguments:

  • Requests research focus from user
  • Classifies intent with confidence scoring
  • Auto-executes high-confidence requests (≥80%)
  • Confirms medium-confidence requests (40-79%)
  • Asks clarifying questions for low-confidence (<40%)

Smart Research - Intelligent Market Research Automation

Purpose: Minimize user prompting while maximizing research autonomy through intelligent intent classification and auto-execution.

How It Works

When user provides research request, this command:

  1. Classifies intent using pattern matching
  2. Calculates confidence in interpretation
  3. Auto-executes or requests minimal clarification
  4. Launches appropriate agents with smart defaults
  5. Provides progress updates and refinement options

Usage Patterns

Auto-Execution (High Confidence ≥80%)

User: "Research Cursor IDE pricing"
System: 🎯 Auto-detected: Pricing analysis for Cursor IDE
🔍 Launching web-search-researcher with pricing focus...
[Executes automatically]

Quick Confirmation (Medium Confidence 40-79%)

User: "Analyze the competitive landscape"
System: 📊 I'll analyze AI IDE competitive landscape including:
• Competitor identification & profiling
• Pricing strategy analysis
• Feature comparison matrix
• Market positioning assessment

Proceed with comprehensive analysis? (y/n/modify scope)

Clarifying Questions (Low Confidence <40%)

User: "Help with my research"
System: 🤔 I can help with several types of research:
[a] Market & competitive analysis
[b] Technical architecture research
[c] Business strategy & positioning
[d] Pricing & business model analysis

What's your primary focus?

Intent Classification Engine

Market Research Patterns

high_confidence_patterns:
"research [company] pricing":
agent: web-search-researcher
scope: pricing_analysis
confidence: 0.95

"analyze [company] vs [company]":
agent: competitive-market-analyst
scope: comparative_analysis
confidence: 0.90

"[company] competitive analysis":
agent: competitive-market-analyst
scope: competitor_profiling
confidence: 0.85

medium_confidence_patterns:
"competitive landscape":
agent: orchestrator
scope: market_intelligence_workflow
confidence: 0.70
clarification: "Focus area? (pricing/features/positioning/comprehensive)"

"market research":
agent: competitive-market-analyst
scope: market_analysis
confidence: 0.60
clarification: "Specific companies or broad market analysis?"

low_confidence_patterns:
"research":
confidence: 0.30
questions: ["What type of research?", "Specific companies/market?", "Business or technical focus?"]

Auto-Scope Detection

Context Clues for Smart Defaults

company_detection:
patterns: ["Cursor", "GitHub Copilot", "Tabnine", "Codeium", "Replit"]
action: auto-focus analysis on detected companies

pricing_keywords:
patterns: ["pricing", "cost", "subscription", "plans", "tiers"]
action: emphasize pricing analysis in research scope

comparison_indicators:
patterns: ["vs", "versus", "compared to", "against"]
action: setup comparative analysis framework

launch_context:
patterns: ["launching", "building", "creating", "developing"]
action: include positioning and differentiation analysis

Execution Logic

def smart_research_execution(user_input):
# Step 1: Intent Classification
intent, confidence, suggested_agent, scope = classify_intent(user_input)

# Step 2: Confidence-Based Routing
if confidence >= 0.8:
return auto_execute(suggested_agent, scope, user_input)
elif confidence >= 0.4:
return quick_confirmation(intent, scope, user_input)
else:
return clarifying_questions(intent, user_input)

def auto_execute(agent, scope, context):
# Launch agent with smart defaults
enhanced_prompt = f"""
Execute {scope} analysis based on: {context}

Auto-detected scope: {scope}
Context clues: {extract_context_clues(context)}
Smart defaults: {generate_smart_defaults(scope, context)}

Provide progress updates and offer refinement options upon completion.
"""

return launch_agent(agent, enhanced_prompt)

Progressive Disclosure Points

Natural Stopping Points

initial_findings_complete:
trigger: "Preliminary research gathered"
action: "Present key findings + offer deep-dive options"
options: ["Focus on top competitors", "Expand to adjacent markets", "Deep-dive pricing", "Technical analysis"]

scope_expansion_opportunity:
trigger: "Interesting adjacent area discovered"
action: "Offer to expand research scope"
example: "Found interesting partnership strategies. Include partnership analysis?"

conflicting_data_detected:
trigger: "Contradictory information found"
action: "Request clarification on approach"
example: "Found conflicting pricing data. Focus on official sources or include analyst estimates?"

Enhanced Agent Integration

Agent Capability Enhancement

Each research agent gets enhanced with:

enhanced_capabilities:
auto_scope_detection: true
context_awareness: true
progress_reporting: true
refinement_suggestions: true

auto_reporting_triggers:
- 25% progress: "Initial data gathering complete"
- 50% progress: "Core analysis underway"
- 75% progress: "Synthesis and insights generation"
- 100% progress: "Analysis complete + refinement options"

Usage Examples

Example 1: Automatic Execution

User: "Research GitHub Copilot pricing for enterprise customers"

System: 🎯 Auto-detected: Enterprise pricing analysis for GitHub Copilot
🔍 Launching web-search-researcher...
📊 Progress: Gathering official pricing data [██████████] 100%
📋 Found: Enterprise pricing tiers, volume discounts, feature comparison

✅ Analysis Complete:
• Enterprise tier: $39/user/month
• Volume discounts: 10% at 100+ users, 20% at 500+ users
• Enterprise features: Advanced security, audit logs, priority support

🔍 Next Steps:
[a] Compare with competitors' enterprise pricing
[b] Analyze ROI justification strategies
[c] Research enterprise sales process

Example 2: Smart Orchestration

User: "I need to understand how CODITECT should position against Cursor"

System: 🎯 Detected: Strategic positioning analysis
📊 This requires multi-faceted analysis. I'll coordinate:
• Competitive analysis (features, pricing, positioning)
• Market positioning assessment
• Differentiation opportunity identification

🚀 Launching orchestrated research workflow...

[Coordinates multiple agents automatically]
[Provides integrated strategic recommendations]

Success Metrics

  • Prompt Reduction: 70% fewer clarification requests needed
  • Time to Insight: 80% faster research initiation
  • Intent Accuracy: 85%+ correct auto-classification
  • User Satisfaction: Preferred automation vs manual agent selection

Smart research command for maximizing automation while ensuring accurate intent fulfillment

Action Policy

<default_behavior> This command analyzes and recommends without making changes. Provides:

  • Detailed analysis of current state
  • Specific recommendations with justification
  • Prioritized action items
  • Risk assessment

User decides which recommendations to implement. </default_behavior>

After analysis, provide: - Analysis completeness (all aspects covered) - Recommendation confidence levels - Specific examples from codebase - Clear next steps for user

Success Output

When smart research completes:

✅ COMMAND COMPLETE: /smart-research
Query: <research-request>
Intent: <detected-intent>
Confidence: XX%
Agents Used: N
Findings: Summary available
Next: [a] Deep dive [b] Export [c] Refine

Completion Checklist

Before marking complete:

  • Intent classified
  • Confidence calculated
  • Appropriate agents launched
  • Findings synthesized
  • Next steps offered

Failure Indicators

This command has FAILED if:

  • ❌ Intent unclassifiable
  • ❌ No agents available
  • ❌ Research timeout
  • ❌ No findings produced

When NOT to Use

Do NOT use when:

  • Need specific codebase research (use /research-codebase)
  • Need biographical research (use /research-bio)
  • Simple fact lookup (use web search directly)

Anti-Patterns (Avoid)

Anti-PatternProblemSolution
Vague requestLow confidenceBe specific
Skip confirmationWrong intentVerify medium-confidence
Ignore refinementIncomplete researchUse progressive disclosure

Principles

This command embodies:

  • #3 Complete Execution - Auto-execute high confidence
  • #5 No Assumptions - Clarify low confidence

Full Standard: CODITECT-STANDARD-AUTOMATION.md