YoungseokOh

Use when running the LGBM autoresearch loop, editing candidate YAML specs, gating new variants, or promoting winners in polymarket-tmax-lab.

Use when working on bootstrap-lab, warehouse management, seed export or restore, legacy raw/parquet cleanup, or canonical data layout in polymarket-tmax-lab.

Use when working on dataset materialization, model training, backtesting, scan-edge signals, paper-trading workflows, or experiment artifacts in polymarket-tmax-lab.

Use when working on bootstrap-lab, warehouse management, seed export or restore, legacy raw/parquet cleanup, or canonical data layout in polymarket-tmax-lab.

Use when working on dataset materialization, model training, backtesting, scan-edge signals, paper-trading workflows, or experiment artifacts in polymarket-tmax-lab.

Use when running the LGBM autoresearch loop, editing candidate YAML specs, gating new variants, or promoting winners in polymarket-tmax-lab.

Use when changing Polymarket discovery, temperature market filtering, rule parsing, settlement source handling, station mapping, or official truth adapters.

Use when validating that polymarket-tmax-lab is experiment-ready after bootstrap, training, backtest, or machine-to-machine seed restore.

Use when working on weekend operations, the queued-job worker, ops-calendar authority, per-GP CLI commands, model promotion gates, paper trading, or operator-assisted live trading for the F1 prediction-market stack.

Use when working on connectors, ingestion jobs, Bronze/Silver/Gold storage, schema migrations, lineage, entity mapping, or data quality for F1 and Polymarket data.

Use when producing EDA summaries, analyst notes, model explanations, data lineage summaries, or concise research reports from structured project data.

Use when preparing a change for handoff, review, release notes, or commit drafting, especially when validation coverage and commit message quality matter.

Use when building features, snapshots, experiments, models, calibration, evaluation, or backtests for the F1 prediction-market research stack.

Use when changing the FastAPI service, Pydantic schemas, TypeScript contracts, Next.js dashboard, or analyst-facing exploration workflows.

Use when building, validating, or debugging KR snapshot bundles and the daily batch flow. Trigger on build_kr_snapshot.py, run_daily_batch.py, snapshot.json, manifest.json, summary.json, GitHub Actions daily-market-data.yml, ci_safe profile behavior, or eit-research bundle loading.

Use when official KRX-backed pykrx paths fail in eit-market-data. Trigger on KRX login changes, `KrxAuthRequired`, `LOGOUT`, `400`, `KeyError('지수명')`, empty KRX index or listing frames, pykrx dataframe column errors like missing `종가`/`BPS`/`PER`, browser cookie/profile issues, or when scripts/krx_login.py and src/eit_market_data/kr/krx_auth.py need to be modified or debugged.

Use before running Korean market-data collection, bundle export, or daily batch commands in eit-market-data. Trigger when the user wants to run or debug preflight, crawl_kr_data_pykrx.py (official raw collection), `crawl_kr_data_fallback.py` (legacy FnGuide repair path), `crawl_kr_data.py` (legacy authenticated recovery), build_kr_snapshot.py, run_daily_batch.py, GitHub Actions data jobs, or local environment setup. Also use when validating API keys, package installation, output directories, or batch readiness.

Use whenever modifying SnapshotBuilder, provider fetch methods, bundle export/load logic, date filters, or any code that could leak future information into snapshots or backtests. Trigger on look-ahead bias concerns, as_of propagation, report_date handling, benchmark joins, sector map timing, or changes to snapshot schemas and validation.

Use when adding or changing market-data providers, provider factories, optional dependencies, scripts that call providers, or tests around DART, ECOS, FDR, pykrx, and snapshot assembly. Trigger on changes under src/eit_market_data/, pyproject optional dependencies, provider protocols, or new market-data fields and adapters.

Guide for analyzing Moltbook AI agent social network data

Generate detailed, solo-developer-focused app/web product ideas from AI-Hub Korea dataset metadata. Use when the user wants to brainstorm app ideas from a dataset, asks "이 데이터셋으로 뭘 만들 수 있어 / 앱 아이디어 뽑아줘", "what can I build with dataset X", or wants to judge feasibility and rank ideas. Always produces at least 5 ideas per dataset with a rich per-idea spec, metadata-only.