What does this skill do, and when should you use it?
This is a Claude Skill that activates automatically whenever a user uploads or references a CSV file. It loads the data with Python and pandas, identifies its structure (sales, customer, financial, operational, or survey data), and produces summary statistics, missing-data analysis, and only the visualizations relevant to that dataset — time-series plots, correlation heatmaps, histograms, and categorical breakdowns. Its defining design principle is proactive: it never asks what you want and instead delivers a complete analysis in one pass. The repo ships an analyze.py engine, dependency manifest, and a bundled demo dataset of 15 months of P&L data with 25 financial metrics.
Reads a CSV file into a pandas DataFrame; identifies column types, date columns, and numeric columns; classifies the business scenario (sales/customer/financial/operational/survey) from the data; runs relevant statistical analysis (trends, correlations, distributions, cross-tabulations); detects and reports missing values; generates only the charts the data supports; and returns a comprehensive text summary plus PNG visualizations via a summarize_csv(file_path) function.
- An e-commerce or sales analyst with an orders CSV who wants immediate revenue trends and product performance
- A finance user uploading transaction or P&L data who needs trend detection and correlation analysis
- A survey researcher with categorical responses and ratings who needs frequency distributions and cross-tabs
- An operations person with timestamped metrics data who wants time-series and performance analysis
- Anyone wanting a fast read on the structure and quality of an unfamiliar CSV inside a Claude conversation
How do you install this skill?
- SKILL.md forces skipping user confirmation; any CSV immediately triggers full analysis and chart generation, which may write/overwrite same-named PNG files locally — watch your output directory.
- Documentation drifts from code: it claims plots only when date columns exist, but distributions and other charts are always generated.
- No LICENSE file in evidence and publisher unverified; verify licensing before enterprise use.
- Date parsing uses errors='coerce'; degenerate inputs can produce NaT-driven runtime errors; large files are unprotected.
- Static review only — code was not executed; all conclusions derive from source reading with low confidence.
- Shell / CLI
- Local filesystem
Python >=3.8pandas >=2.0.0matplotlib >=3.7.0seaborn >=0.12.0
For Claude.ai users: 1) Download csv-data-summarizer.zip from the repo's Releases; 2) Open Claude.ai → Settings → Capabilities → Skills; 3) Upload the zip and enable the skill. For developers: git clone [email protected]:coffeefuelbump/csv-data-summarizer-claude-skill.git && cd csv-data-summarizer-claude-skill && pip install -r requirements.txt.
git clone --depth 1 https://github.com/coffeefuelbump/csv-data-summarizer-claude-skill.git ~/.claude/skills/csv-data-summarizer-claude-skillGenerated from the source repository and skill path; it copies only this skill's folder. If the author's install steps above differ, follow those first. To scope it to one project, replace ~/.claude/skills with that project's .claude/skills.
How do you use this skill?
Once installed, send your agent any of these to trigger it:
- Here's sales_data.csv. Can you summarize this file?
- Analyze this customer data CSV and show me trends.
Upload any CSV file in Claude.ai — the skill activates automatically and runs a full analysis immediately, with no extra prompting. Example triggers: "Here's sales_data.csv. Can you summarize this file?" or "Analyze this customer data CSV and show me trends." Output includes a dataset overview, summary statistics, missing-value checks, correlations, and multiple charts.
What are this skill's strengths and limitations?
- Zero-prompt design: upload triggers immediate analysis with no questions or option menus
- Adaptive analysis tailored to data type (sales, financial, survey, etc.)
- Charts are generated conditionally, only when the data supports them (e.g., time-series only with date columns)
- Bundled 15-month financial P&L demo dataset makes it easy to verify behavior
- Clear dependencies (Python 3.8+, pandas 2.0+, etc.) and a documented example output
- Repo topic is claude-skills and the README claims MIT, but GitHub metadata lists the license as unknown — verify the LICENSE file yourself
- The forced no-questions behavior may produce unwanted output on sensitive or ambiguous datasets
- No test suite is provided and no evidence of validation across platforms
- Internal inconsistency in the skill spec: the Notes section says visualizations are generated only when date columns exist, which contradicts the body's multi-chart description
How does this skill compare with similar options?
Side by side with related skills; every score comes from the same FSRS standard.
| Skill | FS score | Stars | Last updated | License |
|---|---|---|---|---|
| CSV Data Summarizer Skill this page | 45 · Not recommended | ★ 471 | 1y ago | — |
| CSV/Excel Deep Data Analysis Skill | 47 · Use with care | ★ 20k | 14d ago | MIT |
| Nature Figure Making Skill (nature-figure) | 60 · Recommended | ★ 48k | 11d ago | Apache-2.0 |
| Workspace Data Analyst | 54 · Use with care | ★ 11k | 27d ago | MIT |
| OfficeCLI Data Dashboard Skill | 57 · Use with care | ★ 32k | 12d ago | Apache-2.0 |
How did FollowSkills review this skill?
No malicious behavior, no network exfiltration, no credential access; data flow is local CSV read producing text/PNG output, reasonably transparent. However SKILL.md uses strong directives to bypass user confirmation ('DO NOT ASK', execute immediately with all charts), an overreach without consent, with no rollback or file-overwrite warnings; deducted mainly for this.
analyze.py's happy path is self-consistent: loading, stats, correlations, time series, and distributions are implemented, with a fallback for single-numeric-column axes. Inconsistencies remain: SKILL.md Notes claim visualizations only when date columns exist, but the code always produces distribution/correlation charts; date coercion with errors='coerce' can leave all-NaT causing min/max failures; no protection for large files or high-cardinality categoricals; no tests or designed failure feedback; deducted here.
Scenario is clear (CSV analysis) and trigger conditions are explicit; adaptive-by-industry intent is reasonable. But non-fit boundaries (huge files, non-tabular CSV, encodings) are undeclared, no Chinese-language support notes; purely local dependencies so no mainland-China reachability issue; environment-fit evidence remains limited; deducted here.
Has version 2.1.0, dependency declarations, directory structure, example output and sample data — readable and usable. But documentation drifts from code behavior, README is marketing-heavy, frontmatter license is unknown while README claims MIT (no LICENSE file in evidence), and there is no changelog or maintenance ownership; deducted here.
Core task (summary stats, missing values, correlations, visualizations) is well covered and output is directly usable, beating manual analysis; but this is a static review with no execution evidence, and edge inputs may crash the script requiring user debugging, limiting comparative-benefit evidence; capped under the static ceiling and deducted.
Only source code and author-described example output; no test suite, no CI execution evidence, no third-party reproduction; sample outputs are author claims, not independently verifiable; capped under the static ceiling of 5 and deducted.
Open a dimension to read why it scored that way
Evidence confidence：Low — Mostly static review, author material or a limited demo; useful for discovery, not high-risk decisions.
See the full review method →