Guide · Agent skill
How to use the Imaging Data Commons Cancer Imaging agent skill
Updated · written with AI from the project's own documentation and checked against our catalogue data
The imaging-data-commons skill teaches an AI agent how to search and download public cancer imaging data from the National Cancer Institute Imaging Data Commons (IDC), an archive of radiology scans and pathology slide images. It is written for researchers, data scientists and students who want an agent to find collections, filter them and fetch files. After this guide you will know what the skill does, how to install it, how to phrase requests so the agent uses it, and which licence and safety limits apply. If you have never used agent skills: a skill is a folder with a SKILL.md instructions file and supporting reference guides that an agent loads when a task matches its description. Nothing in the skill runs until the agent decides it is relevant to your request.
Imaging Data Commons Cancer ImagingK-Dense-AI/scientific-agent-skills · MIT Download the skill Quick start
- 1Download the ZIP. Download the imaging-data-commons skill ZIP from this page.
- 2Unzip into a skills folder. Unzip the folder into ~/.claude/skills/ for all projects, or into a project's .claude/skills/ folder.
- 3Or upload in the Claude apps. Open Settings, go to Capabilities, then Skills, and upload the ZIP.
- 4Check Python for downloads. If you need downloads or analysis, run python scripts/check_version.py and follow the install command it prints.
- 5Ask a natural question. Describe the cancer type, modality or body site and what you want back, such as a count, licence or download.
- 6Review licences. Check each collection's licence, mostly CC-BY with some CC-NC, before reuse.
What the skill makes the agent do
Once installed, the agent knows how to query IDC collections by cancer type, modality, or body site, count matching records, check licences and citations, generate viewer links, and download DICOM files. The skill covers radiology (CT, MR, PET) and pathology AI datasets, metadata SQL, and visualization.
The skill tells the agent to pick the lightest access route. For simple read-only questions it uses the IDC REST API over curl without installing anything. For downloads and analysis it uses the idc-index Python client.
- Discover collections and count or filter them by cancer type, modality or body site
- Run metadata SQL queries
- Check per-collection licences and citations
- Generate viewer URLs
- Download DICOM files
Install the skill from the ZIP or the Claude apps
Download the skill ZIP from this page and unzip it. For Claude Code, place the unzipped folder in ~/.claude/skills/ to make it available to all projects, or in a project's .claude/skills/ folder to limit it to that project.
In the Claude apps, you can instead upload the ZIP under Settings → Capabilities → Skills.
- Personal install: ~/.claude/skills/
- Project install: .claude/skills/ inside your project
- Claude apps: Settings → Capabilities → Skills
Set up Python for downloads and analysis
Metadata questions do not need any installation. Downloads, pandas work, plotting and pathology tiling do. The skill runs scripts/check_version.py, which checks for idc-index 0.12.5 and, if it is missing, prints the exact install command for the Python interpreter you are running. The script does not install anything itself.
Installing idc-index adds about 77 MB of packaged index data plus pandas, pyarrow and duckdb. Prefer a virtual environment. Confirm the script's 'meets pinned minimum' message before continuing; a launch failure is not a pass.
python scripts/check_version.pyTrigger the skill with natural requests
You do not need to name the skill. Ask for IDC-related work in plain language, mentioning the cancer type, imaging modality or body site. The agent should load the skill when the request involves public cancer imaging, IDC collections, DICOM downloads, or licence and citation questions.
Requests that fit well are concrete about what you want to find or get. Name the cancer, the modality and whether you need files or only a count.
- How many public IDC collections include CT scans of the lung?
- Find pathology slide collections for breast cancer and tell me their licences.
- Download the MR series for this cohort and show me a viewer link.
- What citation should I use for the TCGA-LIHC collection?
Understand the network and script behaviour
The skill uses the network. Public IDC metadata routes and public GCS and AWS S3 download routes need no authentication. Optional BigQuery and Google Healthcare DICOMweb routes need Google Cloud credentials through Application Default Credentials. Citation resolution contacts DOI services.
The skill includes scripts, so review scripts/check_version.py before running it if your environment is restricted. Some collections are CC-NC, so check each collection's licence before commercial use. Data is for research and is not for patient-identifiable or clinical decisions.
- Public routes: no credentials required
- Optional Google routes: require Google Cloud authentication
- Local Python workflows: idc-index 0.12.5 and about 77 MB of index data
Tips for better results
Ask the agent to confirm the IDC data version before relying on results, since the skill reports v24 as the reviewed version and advises rechecking at use time. For large selections, ask for a dry run first. The CLI guide notes that Boolean options take a value, such as --dry-run true, and that an omitted selection errors.
- Give the cancer type, modality and body site together to narrow results early
- Ask for licences and citations alongside any dataset you plan to publish or reuse
- Ask for a dry run before large downloads
- If the IDC MCP server is connected to your session, the agent will use it for discovery
What people use it for
- Counting public CT or MR collections for a specific cancer type before writing a grant proposal
- Building a cohort of breast pathology slides from IDC metadata with SQL
- Pulling licence and citation text for datasets used in a paper
- Generating viewer links to inspect a DICOM series in a browser
- Downloading a version-pinned set of DICOM series for a reproducible analysis script
Frequently asked questions
Is the imaging-data-commons skill free?
The skill is provided under the MIT License. The IDC data itself has individual licences, mostly CC-BY and some CC-NC, which you must respect when using it.
Do I need an account or credentials to use it?
No for public metadata and download routes, which need no authentication. Google credentials are needed only for the optional BigQuery and Google Healthcare DICOMweb routes.
How do I install the imaging data commons skill?
Download the ZIP from this page and unzip it into ~/.claude/skills/ or a project's .claude/skills/ folder. You can also upload the ZIP in the Claude apps under Settings → Capabilities → Skills.
What do I need installed for downloads?
Python 3 with idc-index 0.12.5. Run python scripts/check_version.py to see whether it is present and get the exact install command. The install adds about 77 MB of index data plus pandas, pyarrow and duckdb.
Can I use IDC data for commercial work?
Only after checking each collection's licence, because some collections are non-commercial CC-NC. The skill's licensing guide covers commercial-use questions and citation formats.
Is it safe to use this data for patient decisions?
No. The data is for research and is not intended for patient-identifiable or clinical care decisions.
New agent skills, repos and AI tools — every week
The best new skills and open-source projects, the fastest-growing repos, and one free prompt. Five minutes to read. Unsubscribe anytime.
Our verdict
Use this skill if you are a researcher, data scientist or student who needs public cancer CT, MR, PET or pathology images and wants an agent to search IDC, check licences and fetch files. Pick something else if you need clinical decision support, identifiable patient data, or commercial use without checking each collection's licence.
Download the skill