What problem does it solve?
It prevents silent Stata run failures by scanning logs for errors, warnings, suspicious patterns, and missing indications that results were actually produced.
Core Features & Use Cases
- Mode A — Log Scan: Greps a Stata log for known error codes and common failure patterns such as invalid syntax, missing files, no observations, undefined variables, and suspicious merge outcomes.
- Sample-size funnel checks: Detects whether the log reports sample sizes at restriction steps (e.g., before/after restrictions) to support econometric best practices.
- Last estimation sanity check: Pulls the most recent estimation output so you can spot implausible magnitudes/signs or problematic SE plausibility.
- Use Case: Verify that a do-file truly ran end-to-end (including merges and restrictions) and that the reported coefficients exist in the log rather than being accidentally fabricated or produced from an incomplete run.
Quick Start
Ask Codex to validate a completed regression run by scanning a specific log file at logs/03_analysis_main_regression.log for errors, warnings, merge issues, sample-size funnel reporting, and the latest estimation summary.