Expert data analyst

Skill mehtab78/skills/skills/expert-data-analyst

Data analysis specialist — SQL, metrics definition, exploratory analysis, aggregation, chart selection, statistical sanity checks, CSV/spreadsheet work. Use for any task involving querying, transforming, or interpreting data. Normally invoked by model-router.From its SKILL.md

Install

npx -y skills add mehtab78/skills --skill expert-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

1 stars1 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.

SKILL.md

1.7 KB, 315 tokens by cl100k_base, as published. Nobody here has run it

Data Analyst Expert

Default tier

haiku for extraction/classification/simple aggregation; sonnet for SQL, pipelines, and analysis; experiment design or causal claims → flag ESCALATE: opus.

Decision rules

Look at the actual data before analyzing it: shapes, nulls, duplicates, date ranges. Never trust column names alone.
State denominators. "Up 40%" is meaningless without base counts.
Correlation language stays correlational unless the design supports causal claims.
Show the query/code that produced every number, so results are reproducible.

Output format

Answer — the headline number(s) with denominators
Method — query/code used, runnable as-is
Caveats — data quality issues found, what would change the answer
Chart only if it adds information a sentence can't

Checklist

Row counts sanity-checked at each join/filter step
Nulls and duplicates handled explicitly, not silently
Time zones / date boundaries stated when dates are involved
Numbers in prose match numbers in output exactly
Verified computationally (script), not mentally

Escalation

Data contradicts the user's stated expectation → report the discrepancy plainly; don't massage it.
PII in the dataset → flag to expert-security-reviewer before outputting rows.

Validation

Re-run the final query/script fresh and confirm outputs match what's reported.

What ships with it

Read from the repository

Just SKILL.md. No reference files, no scripts.

Gives 0 of the 12 instructions most data analysis skills give in 315 tokens

Counted across 230 of the 242 authors here whose files we hold, read 2026-09-06

Propose a regression test for each fixed bugin 16 of 230, across 12 files
Name tests after the bug they preventin 14 of 230, across 10 files
Test the API response shape, not the implementationin 14 of 230, across 10 files
Run the test suite before any code reviewin 14 of 230, across 10 files
Force sandbox mode in the test setupin 14 of 230, across 10 files
Write regression tests only for bugs already foundin 14 of 230, across 10 files
Assert sandbox and production paths return the same shapein 14 of 230, across 10 files
Clear stale state when setting an errorin 13 of 230, across 9 files
Keep the whole test suite under one secondin 10 of 230, across 6 files
Run the build type check before code reviewin 10 of 230, across 6 files
Use vectorized operations instead of row iterationin 9 of 230, across 6 files
Start bar chart Y-axes at zeroin 8 of 230, across 7 files

Said here and by no other author read

State denominators with every headline number
Show the query that produced each number
Report answer, method, and caveats
Add charts only when they add information
Sanity-check row counts at each join and filter
Handle nulls and duplicates explicitly

Grouped from the skills themselves: near-identical wordings counted once, and counted by distinct author, so one author publishing three of these counts once. Length counted with cl100k_base; the agent that loads this file may tokenize it differently.