research

name: research description: "Deep research skill - comprehensive information gathering and analysis using multi-model orchestration" trigger: "research"

Research Skill

Overview

A comprehensive research skill that triggers when you say "research something" or similar phrases. Uses multi-model orchestration for deep, thorough information gathering and analysis.

Trigger Phrases

The skill activates when messages contain:

• "research" (as a verb: "research X", "do research on Y")
• "look into"
• "investigate"
• "find information about"
• "analyze" (when combined with research intent)
• "study" (when research-focused)

Architecture

1. Research Pipeline

Research Request → Query Analysis → Multi-Model Orchestration → Synthesis → Report

2. Model Coordination

• Research Phase: gemini-pro (research capability)
• Analysis Phase: claude-sonnet (reasoning + analysis)
• Writing Phase: gpt-4 (creativity + writing)
• Review Phase: claude-sonnet (review)

3. Information Sources

1. Web search (when Brave API available)
2. Academic/scholarly sources
3. Current news (time-sensitive topics)
4. Official documentation
5. Memory system (past research)

Usage

Basic Research

User: "research quantum computing applications"
→ Skill triggers
→ Multi-model orchestration executes
→ Returns comprehensive research report

Complex Research

User: "research the impact of AI on social work education"
→ Skill triggers with complexity analysis
→ Parallel research on multiple aspects
→ Integrated analysis with citations

Research with Constraints

User: "research renewable energy policies in Europe, focus on Germany"
→ Skill triggers with specific scope
→ Targeted research with geographic focus
→ Comparative analysis

Implementation

Core Components

1. research_skill.py - Main skill handler
2. research_pipeline.py - Multi-stage research workflow
3. query_analyzer.py - Parse research intent and scope
4. source_tracker.py - Track information sources and citations

Integration Points

• Multi-Model Orchestrator - Coordinates AI models for research
• Perplexity-Inspired Skill - Query analysis and source citation
• Async Agent Pattern - Long-running research tasks
• Memory System - Store and retrieve research findings

Research Methodology

1. Scope Definition

• Identify research question
• Determine breadth vs. depth
• Set time constraints (current vs. historical)
• Define required sources

2. Information Gathering

• Multi-source collection
• Source credibility assessment
• Time sensitivity consideration
• Cross-verification

3. Analysis & Synthesis

• Pattern identification
• Trend analysis
• Comparative evaluation
• Gap identification

4. Report Generation

• Structured format
• Source citations
• Key findings summary
• Recommendations/next steps

Output Format

Standard Research Report

# Research Report: [Topic]

## Executive Summary
Brief overview of key findings

## Research Methodology
- Scope definition
- Sources used
- Analysis approach

## Key Findings
1. Finding 1 with supporting evidence
2. Finding 2 with supporting evidence
3. Finding 3 with supporting evidence

## Analysis
- Trends and patterns
- Comparative insights
- Implications

## Sources
1. [Source 1] - Type, date, relevance
2. [Source 2] - Type, date, relevance
3. [Source 3] - Type, date, relevance

## Recommendations
- Further research areas
- Practical applications
- Limitations

## Research Metadata
- Date conducted: [timestamp]
- Models used: [list]
- Confidence score: [0.0-1.0]
- Time spent: [duration]

Configuration

Research Depth Levels

1. Quick - Surface-level, 1-2 models, <5 minutes
2. Standard - Comprehensive, 2-3 models, 5-15 minutes
3. Deep - Thorough, 3+ models, 15-30+ minutes
4. Academic - Scholarly, multiple sources, citations, peer-review style

Source Priority

1. Academic - Peer-reviewed journals, conferences
2. Official - Government reports, organizational publications
3. News - Current events, recent developments
4. Web - General information, blogs, documentation

Examples

Example 1: Technology Research

User: "research blockchain scalability solutions"
→ Triggers research skill
→ Uses: gemini-pro (research) → claude-sonnet (analysis) → gpt-4 (writing)
→ Returns: Comprehensive report on Layer 2, sharding, consensus algorithms

Example 2: Social Science Research

User: "research effects of social media on mental health in adolescents"
→ Triggers research skill
→ Uses: gemini-pro (academic sources) → claude-sonnet (analysis) → review
→ Returns: Research with academic citations, statistical analysis

Example 3: Current Events Research

User: "research recent developments in quantum computing 2025"
→ Triggers research skill with time sensitivity
→ Prioritizes recent sources (<6 months)
→ Returns: Current state analysis with timeline

Integration with Existing Workflow

Heartbeat Integration

Research tasks can be scheduled via heartbeat for:

• Daily research briefings
• Weekly topic deep dives
• Ongoing research projects

Memory Integration

• Store research findings in memory system
• Create research topic embeddings
• Enable semantic search across past research
• Build knowledge graph of research topics

Project Integration

• Link research to thesis projects
• Track research progress
• Organize findings by project
• Generate literature reviews

Quality Assurance

Validation Methods

1. Cross-model verification - Multiple models analyze same data
2. Source credibility scoring - Rate sources by authority/recency
3. Consistency checking - Ensure findings are internally consistent
4. Bias detection - Identify potential biases in sources/analysis

Confidence Scoring

• High (0.8-1.0) - Multiple reliable sources, consistent findings
• Medium (0.5-0.8) - Limited sources, some uncertainty
• Low (0.0-0.5) - Inconclusive, needs more research

Limitations

Current Constraints

1. Web search dependency - Requires Brave API for external research
2. Model availability - Limited to configured AI models
3. Time constraints - Deep research requires significant time
4. Source access - Some sources may be paywalled or restricted

Mitigation Strategies

• Use memory system for past research
• Leverage multi-model analysis for depth
• Implement progressive disclosure (quick → deep)
• Cache research findings for similar queries

Future Enhancements

Planned Features

1. Automated literature review - Academic paper analysis
2. Research topic tracking - Monitor developments over time
3. Comparative research - Side-by-side analysis of multiple topics
4. Visual research outputs - Charts, graphs, timelines
5. Collaborative research - Share findings, peer review

Integration Goals

1. Thesis research assistant - Specialized for academic research
2. News monitoring - Daily research briefings on selected topics
3. Knowledge base builder - Automatically expand memory system
4. Research dashboard - Track all research activities and findings

Usage Notes

Best Practices

1. Be specific - Clear research questions yield better results
2. Set scope - Define breadth/depth expectations
3. Consider time - Recent vs. historical information needs
4. Review sources - Always check source credibility

Common Use Cases

• Academic research support
• Market/competitor analysis
• Technology evaluation
• Policy research
• Historical analysis
• Current events monitoring

Quick Start

Basic Usage

Just say "research [topic]" and the skill will automatically:

1. Analyze your query
2. Determine appropriate research depth
3. Coordinate multiple AI models
4. Return comprehensive findings

Advanced Usage

For more control, specify:

• "deep research on [topic]" - More thorough analysis
• "quick research on [topic]" - Surface-level overview
• "academic research on [topic]" - Scholarly focus
• "current research on [topic]" - Recent developments only