imaging-data-commons

Query and download public cancer imaging data from NCI Imaging Data Commons using idc-index. Use for accessing large-scale radiology (CT, MR, PET) and pathology datasets for AI training or research. No authentication required. Query by metadata, visualize in browser, check licenses.

Install with Tessl CLI

npx tessl i github:K-Dense-AI/claude-scientific-skills --skill imaging-data-commons

What are skills?

1.35x

Quality

92%

Does it follow best practices?

Impact

73%

1.35x

Average score across 3 eval scenarios

SKILL.md

Review

Evals

Evaluation results

53%

25%

Commercial AI Dataset Curation from IDC

License checking and citation generation

Criteria

Without context

With context

IDC version verified

60%

100%

Package version check

License field queried

50%

100%

CC BY-NC excluded

60%

70%

collections_index fetched

Cancer/collection join

Exploratory filter query

28%

LIMIT in exploratory queries

Size estimated

80%

100%

Selection saved to CSV

100%

citations_from_selection used

100%

Citation format constant

Without context: $1.6731 · 16m 37s · 71 turns · 77 in / 17,405 out tokens

With context: $0.6305 · 4m 22s · 25 turns · 696 in / 5,340 out tokens

92%

21%

Segmentation Dataset Inventory for Radiology AI Research

Specialized index discovery and schema exploration

Criteria

Without context

With context

IDC version verified

100%

Package version check

100%

indices_overview accessed

100%

seg_index fetched

33%

100%

analysis_results_index fetched

62%

analysis_result_id used

100%

seg_index to index JOIN

100%

LIMIT in exploratory queries

No BigQuery

100%

Results saved to file

100%

Without context: $2.8893 · 15m 44s · 75 turns · 75 in / 37,031 out tokens

With context: $0.7738 · 2m 31s · 32 turns · 71 in / 6,851 out tokens

75%

10%

Reproducible DICOM Download Pipeline for Multi-Site Imaging Study

Reproducible download pipeline with manifest and CLI

Criteria

Without context

With context

IDC version verified

100%

Size estimated

100%

LIMIT in exploratory queries

download_from_selection used

Custom dirTemplate

100%

Batch download loop

series_aws_url for manifest

100%

Manifest format correct

100%

idc download CLI used

100%

Selection saved to CSV

100%

CRDC UUID naming documented

100%

Without context: $1.0125 · 4m 38s · 46 turns · 46 in / 11,901 out tokens

With context: $1.0175 · 2m 49s · 32 turns · 2,521 in / 9,344 out tokens

Evaluated: 3 days ago
Agent: Claude Code
Model: Claude Sonnet 4.6

Table of Contents

Commercial AI Dataset Curation from IDC Segmentation Dataset Inventory for Radiology AI Research Reproducible DICOM Download Pipeline for Multi-Site Imaging Study

Is this your skill?

If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.