cellxgene-census

Query the CZ CELLxGENE Census programmatically for versioned public single-cell and spatial transcriptomics data. Use when you need population-scale cell metadata, gene expression slices, Census summary counts, source H5AD URIs/downloads, embeddings, spatial Census data, or reference atlas comparisons across organisms, tissues, diseases, assays, and cell types. For analyzing your own local single-cell data use scanpy, anndata, or scvi-tools.

Quality

92%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

Securityby

Low

Low-risk findings worth noting

Quality

Content

85%

Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.

A highly actionable, well-structured skill body with clear workflows and genuine progressive disclosure to two real reference files. The main weakness is conciseness: the redundant 'Common Use Cases' section and standard-scanpy walkthrough add tokens without adding Census-specific value.

Suggestions

Remove or collapse the 'Common Use Cases' section — Use Cases 1–4 duplicate the Core Workflow Patterns already shown in sections 2, 3, 5, and 8, adding length without new information.

Trim the standard scanpy pipeline in section 7 (normalize_total, log1p, highly_variable_genes, PCA, neighbors, UMAP) to just the Census-loading step, since Claude already knows the scanpy workflow.

Consider moving the full pattern catalog in 'Core Workflow Patterns' into references/common_patterns.md and keeping only a quick-start example inline, to reduce body length while preserving detail in the bundle.

Dimension	Reasoning	Score
Conciseness	Mostly efficient Census-specific API detail, but the 'Common Use Cases' section duplicates the Core Workflow Patterns (Use Cases 1–4 re-implement sections 2/3/5/8) and section 7 walks through standard scanpy steps (normalize_total/log1p/PCA/UMAP) Claude already knows.	2 / 3
Actionability	Provides fully executable, import-bearing, copy-paste-ready code throughout (open_soma, get_anndata, axis_query, ExperimentDataset, spatial to_spatialdata), plus concrete filter syntax and commands — matching the executable anchor.	3 / 3
Workflow Clarity	Multi-step processes are clearly sequenced with explicit checkpoints ('Estimate Query Size Before Loading' branches to out-of-core processing; 'Two-Step Workflow: Explore Then Query') and a troubleshooting section for error recovery.	3 / 3
Progressive Disclosure	Body is an overview that pushes detail to two real, one-level-deep bundle files (references/census_schema.md, references/common_patterns.md), each clearly signaled with a 'When to read' clause for easy navigation.	3 / 3
	Total	11 / 12 Passed

Description

100%

Based on the skill's description, can an agent find and select it at the right time? Clear, specific descriptions lead to better discovery.

A strong, third-person description that clearly states capabilities, provides explicit 'Use when' triggers with broad natural-language coverage, and disambiguates from neighboring tools via negative guidance. All four dimensions hit the top anchor with no verbosity or fluff.

Dimension	Reasoning	Score
Specificity	Lists multiple concrete capabilities — 'population-scale cell metadata, gene expression slices, Census summary counts, source H5AD URIs/downloads, embeddings, spatial Census data, or reference atlas comparisons' — matching the multiple-specific-actions anchor.	3 / 3
Completeness	Explicitly answers both what ('Query the CZ CELLxGENE Census programmatically for versioned public single-cell and spatial transcriptomics data') and when ('Use when you need population-scale cell metadata...'), with an explicit 'Use when' clause and negative-scope guidance.	3 / 3
Trigger Term Quality	Covers natural terms users would say ('cell metadata, gene expression slices, summary counts, H5AD URIs/downloads, embeddings, spatial Census data' plus organism/tissue/disease/assay/cell type), giving good coverage rather than just 'some relevant keywords'.	3 / 3
Distinctiveness Conflict Risk	Has a clear niche (CZ CELLxGENE Census) with distinct triggers and an explicit redirect for local data ('For analyzing your own local single-cell data use scanpy, anndata, or scvi-tools'), making misfires unlikely.	3 / 3
	Total	12 / 12 Passed

Validation

93%

Warnings & errors only

Checks the skill against the spec for correct structure and formatting. All validation checks must pass before discovery and implementation can be scored.

Validation — 15 / 16 Passed

Validation for skill structure

Criteria	Description	Result
skill_md_line_count	SKILL.md is long (560 lines); consider splitting into references/ and linking	Warning

	Total	15 / 16 Passed

Repository: K-Dense-AI/scientific-agent-skills
Path: skills/cellxgene-census/SKILL.md
Commit: 757b63b

Reviewed: about 12 hours ago

Table of Contents

Discovery Implementation Validation

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.