Csv Analyzer

Analyze CSV/Excel files with natural language. Get statistics, filter rows, find anomalies, generate summaries, and export results. No pandas needed — uses P...

Security Scan

The skill's code and runtime instructions match the stated CSV/Excel analysis purpose, request no credentials or installs, and contain no hidden network calls or unrelated behavior.

Purpose & Capability

Name/description (lightweight CSV analysis) align with the included Python script and SKILL.md. The script implements stats, filtering, top/bottom, anomaly detection, grouping, and export using only the Python stdlib — consistent with the stated purpose. No extraneous env vars, binaries, or config paths are requested.

Instruction Scope

SKILL.md instructs running the included script against local CSV files; the script only reads supplied file paths and writes an optional output CSV. There are no network calls or references to unrelated system files. Minor note: the filter implementation uses eval(...) for numeric comparisons, but the operator token is constrained by a regex to a small set of comparison operators, and values are cast to float for numeric comparisons, so risk is limited. Also, the filter regex restricts column names to \w+ (no spaces/special chars), which is a usability limitation rather than a security mismatch.

Install Mechanism

No install spec; the skill is instruction-only with a single Python script included. Nothing is downloaded or written at install time.

Credentials

The skill requires no environment variables, credentials, or config paths — appropriate for a local CSV analyzer that operates on user-supplied files.

Persistence & Privilege

always is false and the skill does not request persistent/system-wide changes. It does not modify other skills or agent config; standard autonomous invocation settings apply.

Assessment

This skill appears to be what it says: a lightweight, local CSV analyzer. Before installing or running it, consider: (1) it will read any file path you provide — don't point it at system secrets or files you don't want processed; (2) outputs are written as CSV and opening them in spreadsheet software can expose CSV/Excel formula-injection risks if the source data contains formulas (sanitize or inspect exported files before opening in Excel); (3) the filter command limits column names to alphanumeric/underscore (no spaces) and numeric comparisons use eval on a constrained operator token — while that is limited, avoid running it on untrusted or specially-crafted files if you have strong threat concerns; (4) for very large files (>100MB) this script loads into memory and may be slow; consider using a streaming tool or pandas for big datasets. Overall, there are no hidden network endpoints or credential requests, and the code matches the documentation.

Analyze CSV files with simple commands. Get instant statistics, filter data, detect anomalies, and export results — all without pandas or heavy dependencies.

Usage

Quick stats

python3 {baseDir}/scripts/csv_analyze.py stats data.csv

Shows row count, column types, min/max/mean for numeric columns, unique counts for text columns.

Filter rows

python3 {baseDir}/scripts/csv_analyze.py filter data.csv --where "amount>1000" --output big_orders.csv

Top/Bottom N

python3 {baseDir}/scripts/csv_analyze.py top data.csv --column revenue --n 10
python3 {baseDir}/scripts/csv_analyze.py bottom data.csv --column revenue --n 5

Detect anomalies (values outside 2σ)

python3 {baseDir}/scripts/csv_analyze.py anomalies data.csv --column price

Group and aggregate

python3 {baseDir}/scripts/csv_analyze.py group data.csv --by category --agg "sum:amount" "count:id"

Features

• 📊 Automatic column type detection (numeric, date, text)
• 🔍 Flexible filtering with comparison operators
• 📈 Statistical summary (mean, median, std, min, max, percentiles)
• 🚨 Anomaly detection (z-score based)
• 📋 Grouping and aggregation
• 💾 Export filtered/processed results
• 🪶 Zero external dependencies — Python stdlib only (csv module)

Dependencies

None! Uses only Python standard library.

Why Not Pandas?

Pandas is great but:

• Takes 100MB+ RAM just to import
• Overkill for quick analysis tasks
• This skill runs on 2GB RAM servers without issues
• For truly large datasets, the agent can recommend installing pandas

Limitations

• Designed for files up to ~100MB (loads into memory)
• For larger files, use streaming mode or install pandas
• Date parsing is basic (ISO format preferred)