Menza data analyst

Skill riteshkew/yc-skills/skills/menza-data-analyst

Catalog of all 198 YC Winter 2026 companies + 18 working Claude Code Skills inspired by the ones an open-source tool can replicate. Inspired by, not affiliated with.

Install

npx -y skills add riteshkew/yc-skills --skill menza-data-analyst

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

0 stars0 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

What its author says it does

Copied from the file, not written here

Point this skill at a CSV and ask natural-language questions — total, average, min, max, top-N, filtered count, and group-by — and get a Markdown answer with supporting tables. Trigger when a user wants to explore or summarise tabular data without writing code.

SKILL.md

3.0 KB, as published. Nobody here has run it

Workflow

When this skill triggers, follow these steps in order.

Step 1 — Locate the CSV

Check whether the user has provided a CSV path.

If a path is provided, confirm the file exists and has a header row.
If no path is provided, ask: "Please provide the path to your CSV file. See resources/sales.csv for a working example with columns: order_id, region, product, units, revenue, date."
The engine works with any CSV that has no quoted commas — simple comma-separated values only.

Step 2 — Clarify the question

Check whether the user has asked a question.

If a question is provided, confirm it maps to a supported pattern (see list below).
If no question is provided, ask what they want to know about the data.

Step 3 — Run the analysis engine

Execute from the skill root:

node scripts/ask.mjs <path-to-csv> "<question>"

The engine exits 0 on success (including "unsupported" responses) and exits 1 only on file/input errors. Capture stdout. If the process exits non-zero, surface the stderr message.

Step 4 — Present the answer

The engine outputs a Markdown block containing:

The question (echoed)
The detected intent
The computed answer
A supporting table (for top-N and group-by questions)

Present this output directly. For numeric answers, note the units (e.g. USD, count).

Step 5 — Offer follow-up analysis

After presenting the answer, offer one or two follow-up questions the user might ask based on the result. For example:

After a group-by: "Would you like to see the top N within a specific group?"
After a total: "Would you like to break that down by region or product?"

Supported question patterns

The engine handles a deterministic subset of natural-language data questions. Use the engine for these; compute directly for anything else.

Pattern	Example
Row count	"How many rows?"
Filtered count	"How many rows where region is West?"
Sum of a column	"What is the total revenue?"
Average of a column	"What is the average units?"
Max of a column	"What is the highest revenue?"
Min of a column	"What is the lowest units?"
Top N by a column	"Show me the top 5 by revenue"
Group-by aggregation	"sum of revenue by region"

If the question does not match any pattern or references an unknown column, the engine returns a clear "I can't answer that" message (exit 0) — this is not an error.

Example

See examples/input.md for the dataset and example questions, and examples/output.md for the engine's output.

Run the example yourself from the skill root:

cd skills/menza-data-analyst
bash examples/run.sh