data-scout
Skill0 starsChanged 47 days agoMITView on GitHub
Find, evaluate, and download public datasets suitable for a research question (psych/social science, economics/policy, health). Use when a study needs data and the user has not supplied their own.
In this repository6 other files it gives its agents.
- CLAUDE.mdCLAUDE.md
- analysis-planSkill
- literature-scanSkill
- run-analysisSkill
- stats-reviewSkill
- write-paperSkill
The file
We split this skill into the rules it gives an agent
4 rules. About 0 are common, 3 are distinctive, and 1 is very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈3
Distinctive · estimate
This author's own choices. Worth a look.
1
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 33
Search candidate sources. For each candidate, verify variables exist by downloading a sample or reading the codebook — do not shortlist on …
- Very specific: line 35
After GATE 1 approval, download with a script (analysis/00_download.py) so acquisition is reproducible.
- Distinctive: line 40
If merging datasets (e.g., country-year), document join keys and report how many rows dropped.
- Distinctive: line 41
If no adequate data exists, report that honestly with the closest alternatives and what each would compromise.
≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Data Scout
Goal: identify datasets that actually contain the variables needed, then acquire them reproducibly. Never fabricate or simulate data as a substitute (simulation only if the user explicitly requests it, and clearly labeled).
Where to look (by domain)
Cross-domain / easiest first
- Our World in Data: tidy CSVs, e.g.
https://ourworldindata.org/grapher/{slug}.csv(find slug on the chart page). Metadata: same URL with.metadata.json. - World Bank API (keyless):
https://api.worldbank.org/v2/country/all/indicator/{INDICATOR}?format=json&per_page=20000&date=2000:2025. Find indicators viahttps://api.worldbank.org/v2/indicator?format=jsonor web search "world bank indicator {topic}". - Kaggle, OSF (osf.io), Zenodo, Harvard Dataverse — search web, download direct file links.
Psych / social science
- GSS (General Social Survey): downloadable Stata/SPSS files from gss.norc.org;
pd.read_stataworks. - World Values Survey, European Social Survey (registration may be needed — if so, tell the user exactly what to download and where to put it).
- OSF repositories attached to published papers (check lit-scan results for data availability statements).
Economics / policy
- World Bank (above), OECD SDMX API, Eurostat API, IMF. FRED requires a free API key — if needed, ask the user for one.
Health
- NHANES: direct XPT downloads from
wwwn.cdc.gov/Nchs/Nhanes/{cycle}/{FILE}.XPT; read withpd.read_sas(path, format="xport"). Variable docs on the accompanying .htm pages. - WHO GHO OData API:
https://ghoapi.azureedge.net/api/{IndicatorCode}(keyless). - CDC WONDER, GBD results tool (web download).
Procedure
- From
idea.md, list required variables: outcome, predictor(s), plausible confounders, unit of analysis, time span. - Search candidate sources. For each candidate, verify variables exist by downloading a sample or reading the codebook — do not shortlist on name alone.
- Produce a shortlist (2–4 candidates) in
data/candidates.md: source, N, unit, years, exact variable names, license, access method, limitations. Recommend one. - After GATE 1 approval, download with a script (
analysis/00_download.py) so acquisition is reproducible. Save originals untouched indata/raw/. - Log in
data/provenance.md: URL, query/parameters, download date, file hash (sha256sum), license/terms, and every transformation applied later.
Rules
- Respect licenses and terms; note any citation requirements for the dataset.
- If merging datasets (e.g., country-year), document join keys and report how many rows dropped.
- If no adequate data exists, report that honestly with the closest alternatives and what each would compromise.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

