OpenParable
Skip to content
← Back to skills

Data Audit

Testing & Quality
data-audit

Data Audit is an Agent Skill for testing & quality published in the brycewang-stanford/Auto-Empirical-Research-Skills repository on GitHub: a SKILL.md instruction file an AI agent can load to do this task.

Scans notebooks for data file references and verifies each file exists on disk. Use when checking for broken data paths.

View on GitHub ↗
QUICK START

How to use this skill

Bring this guide into your coding agent with a prompt tailored to the tool you use.

  1. Open your project in Codex.
  2. Copy and paste the prompt below.
  3. Review files and risks before approving any action.
Prompt to paste
Inspect the Agent Skill at https://github.com/brycewang-stanford/Auto-Empirical-Research-Skills/blob/HEAD/skills/29-quarcs-lab-project20XXy/dot-claude/skills/data-audit/SKILL.md and all supporting files. Report provenance, permissions, scripts, dependencies, and security risks. Propose an exact installation into .agents/skills/data-audit, but wait for my approval before writing files or running scripts.

Copying this prompt does not install or run the skill. Review third-party files before use.

Audit Data References

Scan all notebooks for data file references and verify they exist on disk.

Steps

  1. Scan all .ipynb files in notebooks/ for data loading patterns:

    • Python: pd.read_csv(...), pd.read_stata(...), pd.read_excel(...), pd.read_parquet(...), open(...), np.loadtxt(...)
    • R: read.csv(...), read_csv(...), read.dta(...), haven::read_dta(...), readxl::read_excel(...), load(...)
    • Stata: use "...", import delimited "...", import excel "...", insheet using "..."
    • Also check the .md Jupytext pairs for the same patterns
  2. Extract every referenced file path and normalize it:

    • Resolve relative paths from the notebook's directory (notebooks/)
    • Resolve paths using DATA_DIR, RAW_DATA_DIR from config.py / config.R
  3. Check that each referenced file exists in data/rawData/ or data/

  4. Scan data/rawData/ and data/ for all data files present on disk

  5. Report three categories:

    Resolved — referenced and found:

    • File path, which notebook references it, line/cell number

    Broken — referenced but not found:

    • File path as written in code, which notebook, suggested fix (closest matching file, or note that it may need to be downloaded)

    Undocumented — on disk but never referenced by any notebook:

    • File path in data/rawData/ or data/ that no notebook loads
  6. Print a summary: total references, resolved, broken, undocumented files

Error handling

  • If no notebooks exist, report "No notebooks found" and stop.
  • If data/rawData/ does not exist, warn but continue checking data/.