Dataset Scouting

use this whenever the user is examining public data, asking whether a study's metadata or raw data is actually usable, looking for exact raw-data download links, extracting the preprocessing or technical details a paper reports, building or updating a candidate list / shortlist / inclusion table, deciding which accessions are worth the download

9 stars 1 forksResearch & Science Plugin marketplace

Install

Claude Code plugin
/plugin marketplace add HelmholtzAI-Consultants-Munich/helmholtz-agent-skills

The repo is a Claude Code plugin marketplace. Run /plugin afterwards to pick the plugin that includes this skill.

skills CLI (Claude Code, Cursor, Codex and more)
npx skills add HelmholtzAI-Consultants-Munich/helmholtz-agent-skills --skill dataset-scouting
Manual install
git clone --depth 1 https://github.com/HelmholtzAI-Consultants-Munich/helmholtz-agent-skills.git /tmp/helmholtz-agent-skills
mkdir -p ~/.claude/skills
cp -r /tmp/helmholtz-agent-skills/plugins/method-skills/skills/dataset-scouting ~/.claude/skills/dataset-scouting

Personal skills live in ~/.claude/skills. Use .claude/skills in a project folder to share a skill with your team.

From the SKILL.md

# Dataset Scouting

Help the user curate and decide which public datasets they can actually use. Detailed scouting produces a **datasheet** — an Excel workbook plus its Markdown working files, one sourced entry per candidate, available and missing data or metadata, actionable download routes, selected technical details, and a recommendation with its reason. A **candidate** is one dataset the user might acquire (typically, a dataset coming from the same study sharing metadata schema, preprocessing pipeline, technology, etc.).

**The human decides.** You screen, propose and record. Accepting, rejecting, prioritising and emailing authors are the user's calls, and their words in the datasheet are theirs to edit, never yours. Acquiring, loading, harmonizing or mapping a dataset is not this skill. 

## Workflow

Copy this checklist and track progress.

```
Dataset scouting:
- [ ] Step 1: Quick check to understand the domain and potential criteria
- [ ] Step 2: Settle criteria together with the user
- [ ] Step 3: Collect
- [ ] Step 4: Preliminary results presentation and additional alignment
- [ ] Step 5: Double-check pass
- [ ] Step 6: Hand over
```

**Step 1: Quick check to understand the domain and potential criteria**

Reconnaissance, not screening. Do not write candidate files yet.
Read the full skill on GitHub

Details

Skill name
dataset-scouting
Path
plugins/method-skills/skills/dataset-scouting/SKILL.md
License
Apache-2.0
Repo updated
26 October 2026

FAQ

What does the Dataset Scouting skill do?

use this whenever the user is examining public data, asking whether a study's metadata or raw data is actually usable, looking for exact raw-data download links, extracting the preprocessing or technical details a paper reports, building or updating a candidate list / shortlist / inclusion table, deciding which accessions are worth the download

How do I install the Dataset Scouting skill?

Claude Code plugin: /plugin marketplace add HelmholtzAI-Consultants-Munich/helmholtz-agent-skills. Skills go in ~/.claude/skills (personal) or .claude/skills (project), and Claude loads them automatically when a task matches the description.

Is Dataset Scouting popular?

It lives in HelmholtzAI-Consultants-Munich/helmholtz-agent-skills, which has 9 GitHub stars and 1 forks.

Related skills

Grilling

9

HelmholtzAI-Consultants-Munich/helmholtz-agent-skills

Stress-tests plans, decisions, and ideas through probing questions and co-creation. Use when the user asks to be grilled or wants to develop and challenge their thinking through an interview.

Testing & QA

Pureclip Optimization

9

HelmholtzAI-Consultants-Munich/helmholtz-agent-skills

Tune PureCLIP parameters for protein-RNA crosslink site calling on eCLIP, iCLIP or PAR-CLIP data

Other

Openeuler Research

9

processmission/agent-skills

使用 openEuler 权威来源查询发行版、官方文档、社区信息、仓库与软件包,并调研内核、Embedded BSP、设备兼容性和功能支持情况。

Research & Science

Proof Engine Wiki

9

yaniv-golan/proof-engine

Attach Proof Engine proofs to LLM-wiki claims. Use for ingest (extract claims, look up or commission proofs, rewrite pages with badges) and lint (re-verify citations, detect contradictions, report stale proofs) steps in wiki pipelines. Trigger phrases: "ingest wiki page", "lint wiki", "{{prove:}} markers", "verify wiki citations".

Research & Science

Repo Wiki

9

boweneos/repo-wiki-agent-skills

Generate and maintain citation-backed repository wikis as MkDocs-compatible Markdown. Creates baseline docs, detects changes, and incrementally updates only affected pages with file path and line number citations.

Research & Science

Organize Research Literature

9

mao-yifan/organize-research-literature

Batch organize and deeply review local PDF research papers for any research topic into an existing Excel literature-summary workbook. Use when Codex must recursively scan local PDFs offline, apply a user-provided or inferred research profile, extract metadata and body evidence, deduplicate against Excel, preserve workbook formatting, append new literature rows, fill blank cells only, process only newly added PDFs, restrict to high-relevance papers, write Chinese or user-specified academic summaries, translate English residues in summary columns, mark relevance/read-priority labels with colors,

Research & Science

Browse the most popular Claude skills, all research & science skills, or write your own with the skills generator.