CtrlK
BlogDocsLog inGet started
Tessl Logo
SkillPublic

cellxgene-census

Programmatically query the CZ CELLxGENE Census (61M+ cells) when you need cross-tissue, disease, or cell-type expression data for population-scale queries and reference atlas comparisons.

Quality

83%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide
SecuritybySnyk

Passed

No findings from the security scan

SKILL.md
Quality
Evals
Security

Quality

Content

78%Weight 40%Scale 1-5

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

A highly actionable body with an excellent executable example and sound decision guidance for choosing access patterns. Its structural weakness is that two provided reference files (census_schema.md, common_patterns.md) are never linked, leaving the skill a self-contained overview that duplicates reference material instead of navigating to it.

Suggestions

Add a clearly signaled reference section linking [references/census_schema.md] and [references/common_patterns.md] (e.g., under 'Implementation Details': 'Full data schema: See census_schema.md'), and trim the inlined data-model/filtering details that those files already cover.

Drop or fold the 'Key Features' section into 'When to Use' and state the is_primary_data best practice once to remove redundancy and save tokens.

Add a brief verification checkpoint in the example (e.g., sanity-check returned cell counts against census_info summary counts) to give the workflow an explicit validation step.

DimensionReasoningScore

Conciseness

The body is dense with non-generic, skill-specific guidance (install commands, a complete runnable script, filter semantics, access-pattern selection) and does not explain background concepts Claude already knows. Minor trims are available: the 'Key Features' section substantially restates 'When to Use' and 'Implementation Details', and the is_primary_data best practice is stated twice ("always filter primary data" in the example comment and again under 'Filtering semantics'). This fits anchor 4 ("Efficient; minor instances of over-explanation that could be trimmed"); it is not 5 because of that redundancy and not 3 because the padding is genuinely minor rather than a section's worth of unnecessary explanation.

4 / 5

Actionability

The body provides a complete, executable script covering the four core access patterns (open_soma with a pinned version, get_obs metadata exploration, get_anndata in-memory slice, axis_query out-of-core iteration with soma_data/soma_dim_0/soma_dim_1 semantics), plus copy-paste install commands including the optional extras variant. This matches anchor 5 ("Fully executable; copy-paste ready code or commands; specific examples cover the common cases"); it is not 4 because there are no gaps requiring the reader to fill in details for the common cases — the one intentionally brief area (experimental ML) is explicitly flagged as optional.

5 / 5

Workflow Clarity

The sequence is clear: pin a version, open with a context manager, explore metadata, then choose an access pattern — with an explicit decision rule ("< ~100k cells" heuristic choosing get_anndata vs axis_query) that serves as a checkpoint. This fits anchor 4 ("Clear sequence with most checkpoints present; minor validation gaps"); it is not 5 because there are no verification or error-recovery steps (e.g., how to sanity-check cell counts or handle an invalid filter), and it is not 3 because the operations are read-only queries, so the destructive/batch validation cap does not apply and the sequence plus decision checkpoint is explicit rather than implicit.

4 / 5

Progressive Disclosure

The bundle contains references/census_schema.md and references/common_patterns.md, but the body never mentions or links either file — navigation to them is absent while the 'Implementation Details' section (data model, filtering semantics, chunk schema) partially duplicates content that belongs in those references. Per the guideline to score against the actual bundle structure, this matches anchor 3 ("references present but not clearly signaled; content that should be separate is inline"); it is not 4/5 precisely because well-organized bundle references exist but are undiscoverable from SKILL.md, and it is not 2 because the body itself is well-sectioned rather than a monolithic dump.

3 / 5

Total

16

/

20

Passed

Description

82%Weight 40%Scale 1-5

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 description: concrete domain, explicit 'when you need...' trigger clause, and a distinct named resource. Its main weakness is a thin action vocabulary (only "query") and omission of the most common natural terms ("single-cell", "scRNA-seq", "gene expression").

Suggestions

Enumerate 2-3 distinct concrete actions (e.g., 'explore cell/gene metadata, load AnnData slices, run out-of-core queries over millions of cells') instead of the single action 'query'.

Add high-frequency natural terms such as 'single-cell', 'scRNA-seq', and 'gene expression' to the trigger phrase list so users phrasing requests with those words will match.

DimensionReasoningScore

Specificity

The description names the domain concretely ("Programmatically query the CZ CELLxGENE Census (61M+ cells)") but offers essentially one action — "query" — with the rest describing data flavors ("cross-tissue, disease, or cell-type expression data", "population-scale queries", "reference atlas comparisons") rather than distinct skill actions. This matches anchor 3 ("Names domain and 1-2 concrete actions, but not comprehensive"); it is not 4 because there are not several distinct actions (e.g., retrieve, iterate, summarize, train) and not 2 because the domain and one concrete action are clearly stated with a meaningful data scope.

3 / 5

Completeness

It explicitly answers both questions: what — "Programmatically query the CZ CELLxGENE Census (61M+ cells)" — and when — "when you need cross-tissue, disease, or cell-type expression data for population-scale queries and reference atlas comparisons", an explicit trigger clause with concrete trigger phrases. This matches anchor 5; it is not 4 because the 'when' is not merely serviceable but enumerates specific triggering needs, and not below 5 because neither what nor when is vague or implicit.

5 / 5

Trigger Term Quality

Good natural keyword coverage: "cross-tissue", "disease", "cell-type expression data", "population-scale queries", "reference atlas" — terms a bioinformatics user would plausibly say, anchored by the recognizable product name "CZ CELLxGENE Census". A few common terms are missing, most notably "single-cell", "scRNA-seq"/"RNA-seq", and "gene expression", so it fits anchor 4 rather than 5's comprehensive synonym coverage; it clears anchor 3 because multiple relevant natural phrases are present, not just one generic keyword.

4 / 5

Distinctiveness Conflict Risk

"CZ CELLxGENE Census" is a specific named external resource with a clear niche (population-scale single-cell atlas queries), and the trigger needs (cross-tissue/disease atlas comparisons) are unlikely to fire for unrelated skills. This matches anchor 5 ("Clear niche with distinct triggers; minimal conflict risk"); at most it could brush against a generic single-cell analysis skill, which would suggest 4, but the named-resource specificity keeps conflict risk minimal.

5 / 5

Total

17

/

20

Passed

Validation

93%

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

CriteriaDescriptionResult

frontmatter_unknown_keys

Unknown frontmatter key(s) found; consider removing or moving to metadata

Warning

Total

15

/

16

Passed

Repository

aipoch/medical-research-skills

Path

scientific-skills/Evidence Insight/cellxgene-census/SKILL.md

Reviewed

Table of Contents

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.