app-sensing-market-research

name	app-sensing-market-research
description	Collect app-store evidence for product discovery and let Codex write the final analysis. Use when Codex needs to handle 시장 조사, 키워드 조사, 경쟁앱 조사, 리뷰 근거 수집, opportunity scoring, or report/email automation for app ideas without calling an external LLM.

App Market Research

Use this skill to collect store evidence first and reason second.

Workflow

1. Run ./scripts/bootstrap_runtime.sh if dependencies are not installed.
2. Run node scripts/market_research.mjs config init once per workspace.
3. Run node scripts/market_research.mjs collect ... to create research_bundle.json.
4. Read references/research-bundle.schema.json and the generated bundle.
5. Write analysis_report.json and analysis_report.md yourself. Do not call an external LLM.
6. If the user wants delivery, run node scripts/market_research.mjs email --report-json ... or notify --report-json ....

Commands

• node scripts/market_research.mjs collect --seed "AI,PDF"
• node scripts/market_research.mjs collect --keyword "AI 가계부"
• node scripts/market_research.mjs collect --batch "AI 가계부,습관 추적기"
• node scripts/market_research.mjs config show
• node scripts/market_research.mjs email --report-json output/sensing/<run_id>/analysis_report.json
• node scripts/market_research.mjs notify --report-json output/sensing/<run_id>/analysis_report.json

Operating Rules

• Treat research_bundle.json as evidence, not the final answer.
• Use the bundle's top keywords, competitor data, and low-rating excerpts to write the report.
• Keep conclusions tied to explicit evidence in the bundle.
• Do not invent review or competitor data that is not present in the files.
• Use references/automation.md when the user wants recurring execution.

References

• references/setup.md
• references/usage-recipes.md
• references/research-bundle.schema.json
• references/analysis-report.schema.json
• references/automation.md