Collaborative Research Workflow

Core Principle: Divide research into Lucky (data gathering) + Jinx (analysis) for maximum efficiency and parallel processing.

When to Use This Skill

✅ Perfect for:

• Market research (competitor analysis, pricing)
• API documentation review
• Trend analysis (Google Trends, marketplaces)
• Technical documentation analysis
• Large-scale content analysis
• Multi-source data comparison

❌ Not suitable for:

• Simple lookups (use direct web_search/web_fetch)
• Real-time data that changes quickly
• Single-page analysis (not worth the overhead)

The 3-Phase Process

Phase 1: Raw Data Gathering (Lucky)

Time: 30-60% of total project time

Focus: Speed and coverage, not precision

1. Set up data directory structure

```bash

mkdir -p /workspace/research/raw-data/YYYY-MM-DD-project

```

1. Use Puppeteer for systematic data collection
• Navigate to target sites
• Capture BOTH html and text: { html: document.body.innerHTML, text: document.body.innerText }
• Save with metadata: URL, timestamp, query/source
• Don't fight DOM selectors — capture everything

1. Save structured files for Jinx

```

METADATA:

URL: [source_url]

TIMESTAMP: [iso_timestamp]

QUERY: [search_query]

RAW TEXT:

[page_text_content]

RAW HTML:

[full_html_content]

```

1. Transfer to Mac Mini SSD

```bash

scp -i ~/.ssh/lucky_to_mac file.html luckyai@100.90.7.148:~/temp/

ssh -i ~/.ssh/lucky_to_mac luckyai@100.90.7.148 "mv ~/temp/* '/Volumes/Crucial X10/research/raw-data/project/'"

```

Phase 2: Parallel Analysis (Jinx)

Time: 20-40% of total project time

Focus: Pattern extraction and structured output

1. Task Assignment Validation
• ✅ Analyzing local files (no internet needed)
• ✅ Structured data processing
• ✅ Text analysis and extraction

1. Send structured analysis tasks to Jinx

```bash

curl -X POST http://localhost:3001/task -H 'Content-Type: application/json' -d '{

"prompt": "Analyze files in /Volumes/Crucial X10/research/raw-data/project/. Extract: [specific_data_points]. Output structured JSON with [required_format]. Provide analysis summary with [specific_insights].",

"priority": "high"

}'

```

1. Key prompting strategies for Jinx:
• Be specific about data extraction requirements
• Request JSON output format
• Ask for both raw findings AND summary analysis
• Include comparison requirements if multiple sources

Phase 3: Compilation & Skills Documentation (Lucky)

Time: 10-20% of total project time

Focus: Synthesis and actionable insights

1. Collect Jinx results

```bash

curl -s http://localhost:3001/results/[task-id]

```

1. Compile comprehensive report
• Executive summary with key findings
• Structured data tables/comparisons
• Strategic recommendations
• Process insights and improvements

1. Document process learnings
• What worked well / areas for improvement
• Time saved vs sequential approach
• Quality of analysis vs manual extraction

Best Practices

Data Gathering (Lucky)

• Capture everything — let Jinx filter, don't pre-filter
• Use consistent file naming — project-source-timestamp.html
• Include rich metadata — helps Jinx understand context
• Work in batches — send first batch to Jinx while gathering more

Analysis Tasks (Jinx)

• Be specific about extraction requirements
• Request execution — ask Jinx to run analysis scripts, not just provide them
• Structure output — JSON format for easy parsing
• Ask for insights — not just data extraction but pattern analysis

Collaboration

• Send tasks early — don't wait for all data before starting analysis
• Check progress regularly — curl status API to monitor queue
• Quality over quantity — better to analyze fewer sources deeply

Time Estimates

Effective time = max(Lucky, Jinx) due to parallelization

Security Considerations

• HTML sanitization — Strip

  Skill工具集 © 2026