| name | tigrimos-market-research |
| description | Multi-agent market research workflow for TigrimOS with brand discovery, data analysis, charts, and compiled report |
TigrimOS Market Research Pipeline
Overview
This skill orchestrates a multi-agent workflow to perform comprehensive market research for TigrimOS (or similar AI assistant/agent platforms). It handles brand name correction, market data collection, visualization generation, and report compilation.
Multi-Agent Topology
[Orchestrator]
├── spawn: brand_discovery_agent (sequential, waits)
├── spawn: market_researcher (parallel)
├── spawn: data_analyst (parallel)
├── spawn: report_compiler (parallel, waits for analyst)
└── synthesize: all outputs → final report
Agent Roles
| Agent | Label | Role | Waits For |
|---|
| Orchestrator | orchestrator | Coordinates workflow, synthesizes results | - |
| Brand Discovery | brand_discovery_agent | Typos/variation handling | orchestrator |
| Market Researcher | market_researcher | Web search, data gathering | brand_discovery |
| Data Analyst | data_analyst | Charts, visualizations, JSON data | market_researcher |
| Report Compiler | report_compiler | Markdown report assembly | market_researcher, data_analyst |
Skills Loaded Per Agent
- orchestrator: No specific skill load
- brand_discovery_agent: Load
brand-discovery skill
- market_researcher: No specific skill load
- data_analyst: Load
language-ranking-chart skill
- report_compiler: No specific skill load
Workflow
Phase 1: Brand Discovery
- Accept user query (may contain typos)
- Load
brand-discovery skill
- Generate spelling variations and search iteratively
- Confirm correct brand name (e.g., "TigrimOS")
- Pass verified name to downstream agents
Phase 2: Market Research (Parallel Agents)
- Market Researcher: Search web for:
- Market size and growth projections
- Competitive landscape (OpenAI, Anthropic, Google, Microsoft)
- Target audience segments
- SWOT analysis
- Data Analyst: Generate visualizations:
- Market growth chart (line/bar)
- Competitive landscape chart (bar)
- Target audience distribution (pie)
- SWOT analysis diagram
- Save as PNG files with JSON data backup
Phase 3: Report Compilation
- Load market research data
- Load chart outputs
- Generate comprehensive markdown report:
- Executive summary
- Market overview with tables
- Charts and visualizations
- Strategic recommendations
- Save to output directory
Implementation
def tigrimos_market_research_pipeline(query: str) -> dict:
"""Execute multi-agent market research workflow."""
brand_agent = spawn_agent("brand_discovery_agent",
skills=["brand-discovery"])
correct_name = brand_agent.discover(query)
market_researcher = spawn_agent("market_researcher")
data_analyst = spawn_agent("data_analyst",
skills=["language-ranking-chart"])
market_data = market_researcher.research(correct_name)
charts = data_analyst.create_charts(market_data)
report_compiler = spawn_agent("report_compiler")
final_report = report_compiler.compile(
product_name=correct_name,
market_data=market_data,
charts=charts
)
return {
"product_name": correct_name,
"market_data": market_data,
"charts": charts,
"report": final_report
}
Expected Outputs
| Output | File | Agent |
|---|
| Market research data | *_market_research_report.md | market_researcher |
| Growth chart | market_growth_chart.png | data_analyst |
| Competitive landscape | competitive_landscape.png | data_analyst |
| Target audience | target_audience.png | data_analyst |
| SWOT analysis | swot_analysis.png | data_analyst |
| Final report | *_Final_Report.md | report_compiler |
Best Practices
- Always run brand discovery FIRST to handle typos ("tigrimos" → "TigrimOS")
- Run market research and data analysis in parallel for efficiency
- Data analyst should wait for market researcher results before creating charts
- Report compiler should wait for both research and charts
- Save all charts as PNG + JSON for downstream use
- Use descriptive filenames with product name prefix
Error Handling
- If brand discovery finds no match: Ask user to verify spelling
- If market researcher times out: Retry once, then fall back to direct web search
- If data analyst times out: Skip visualizations, continue with text-only report
- If report compiler times out: Use orchestrator to directly synthesize results
Output Directory Structure
./sandbox/output_file/
├── TigrimOS_Marketing_Research_Report.md
├── market_growth_chart.png
├── competitive_landscape.png
├── target_audience.png
├── swot_analysis.png
└── [data_analyst_outputs.json]