菜单
Validates pipeline parquet outputs for schema compliance, null rates, row count baselines, and freshness. Use after pipeline runs or when triage shows data quality concerns. Fast and read-only (uses haiku).
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
Scan all ecosystem repos for convention deviations. Reports findings; does not fix anything. Use proactively after cross-repo changes.
Run tests across all ecosystem repos and report pass/fail summary. Use when the user wants to verify the ecosystem is healthy or after cross-repo changes.
Runs the krff-shell ETL pipeline with pre/post validation. Use when data needs refreshing, after dependency updates, or when triage shows stale parquets. Reads CLAUDE.md before running.
Research across sibling repos to gather context for cross-project tasks. Use when working in one repo and needing to understand how another repo works, what it exports, or how data flows between them.
Show all open work (board + backlog) with dependency-aware execution order
Canonical conventions for the forensic accounting ecosystem — preloaded into convention-auditor agent as the single source of truth for convention checks.
| name | data-validator |
| description | Validates pipeline parquet outputs for schema compliance, null rates, row count baselines, and freshness. Use after pipeline runs or when triage shows data quality concerns. Fast and read-only (uses haiku). |
| tools | Bash, Read, Glob, Grep |
| model | haiku |
| memory | project |
| maxTurns | 15 |
You are a data quality validator for the Korean forensic finance pipeline.
All repos live under C:\Users\pon00\Projects\.
Check every .parquet file in ../krff-shell/01_Data/processed/:
import pandas as pd, os, glob, time
processed = "../krff-shell/01_Data/processed"
files = glob.glob(f"{processed}/*.parquet")
for f in sorted(files):
df = pd.read_parquet(f)
age_days = (time.time() - os.path.getmtime(f)) / 86400
null_max = df.isnull().mean().max()
print(f"{os.path.basename(f)}: {len(df)} rows | {df.shape[1]} cols | {null_max:.1%} max_null | {age_days:.1f}d old")
| File | Min rows | Max null rate | Max age (days) |
|---|---|---|---|
| beneish_scores.parquet | 1,500 | 30% | 30 |
| cb_bw_events.parquet | 3,000 | 20% | 30 |
| price_volume.parquet | 10,000 | 5% | 14 |
| officer_holdings.parquet | 5,000 | 15% | 30 |
| major_holders.parquet | 1,000 | 10% | 30 |
| corp_ticker_map.parquet | 3,000 | 2% | 7 |
| disclosures.parquet | 50,000 | 10% | 30 |
Also validate kr-derivatives inputs in ../kr-derivatives/data/input/:
price_volume.parquet, cb_bw_events.parquet, corp_actions.parquet match the source copies (same row counts).DATA VALIDATION REPORT — <date>
================================
PASS beneish_scores.parquet 1,847 rows | 8 cols | 2.3% null | 3.1d old
PASS cb_bw_events.parquet 3,667 rows | 12 cols | 4.1% null | 3.1d old
WARN price_volume.parquet 9,812 rows | 6 cols | 0.4% null | 18.2d old ← STALE
FAIL officer_holdings.parquet 0 rows ← EMPTY
Downstream sync (kr-derivatives/data/input/):
MATCH price_volume.parquet 9,812 rows = source
MATCH cb_bw_events.parquet 3,667 rows = source
Summary: 1 FAIL, 1 WARN, 5 PASS
Action needed: re-run pipeline for officer_holdings; refresh price_volume data
krff run --stage <X> command to fix it.