anndata

Data structure for annotated matrices in single-cell analysis. Use when working with .h5ad files or integrating with the scverse ecosystem. This is the data format skill—for analysis workflows use scanpy; for probabilistic models use scvi-tools; for population-scale queries use cellxgene-census.

Quality

81%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

Securityby

Passed

No findings from the security scan

Quality

Content

72%

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

The body is highly actionable with well-structured progressive disclosure via real reference files. It loses points on conciseness (repeated examples across sections) and workflow clarity (batch/destructive operations lack validation checkpoints).

Suggestions

De-duplicate the backed-mode, strings_to_categoricals(), and compressed-write examples so each appears once, cross-referencing the relevant reference file instead of restating the snippet.

Add an explicit verification step after batch concatenation and after QC filtering (e.g. 'print(adata)' / assert shape / check .obs['batch'].value_counts()) to satisfy the validation-checkpoint requirement and lift workflow_clarity to 3.

Move the repeated 'Common commands' blocks out of Core Capabilities into the referenced files, keeping the overview purely navigational.

Dimension	Reasoning	Score
Conciseness	Backed-mode reads, strings_to_categoricals(), and read_h5ad-with-compression each recur across Quick Start, Core Capabilities, Best Practices, and Troubleshooting, so while mostly efficient it carries repetition that could be tightened — the 'mostly efficient but includes some unnecessary explanation or could be tightened' anchor rather than 'every token earns its place' (3).	2 / 3
Actionability	Quotes fully executable blocks such as 'uv pip install anndata', 'ad.concat([adata1, adata2, adata3], axis=0, join="inner", label="batch")', and the scanpy preprocessing pipeline — copy-paste ready, specific examples throughout.	3 / 3
Workflow Clarity	The 'Single-cell RNA-seq analysis' workflow is numbered 1–5 and batch/large-dataset flows are sequenced, but batch concatenation (ad.concat) and destructive filtering (adata = adata[adata.obs['n_genes'] > 200]) have no validation/verification checkpoints, which the rubric caps at 2 for batch/destructive operations.	2 / 3
Progressive Disclosure	Quotes well-signaled one-level-deep pointers ('See: references/data_structure.md', 'references/io_operations.md', 'references/concatenation.md', 'references/manipulation.md', 'references/best_practices.md'), all of which exist as real files, with detail appropriately split out of the overview.	3 / 3
	Total	10 / 12 Passed

Description

90%

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

The description is strong: it states what the skill is, when to use it via an explicit 'Use when' clause, and explicitly disambiguates from three sibling scverse skills. Its only gap is that it describes the data structure's role rather than enumerating concrete actions.

Suggestions

Add concrete verbs to the 'what' clause, e.g. 'Create, read, write, subset, and concatenate annotated matrices (.h5ad) for single-cell analysis.' to lift specificity from 2 to 3.

Consider including the package name 'AnnData' as an explicit trigger term, since users often request it by name.

Dimension	Reasoning	Score
Specificity	Quotes 'Data structure for annotated matrices in single-cell analysis' — it names the domain and the object's role but does not list multiple concrete actions (e.g. create, read, write, concatenate AnnData objects), matching the 'names domain and some actions' anchor rather than 'lists multiple specific concrete actions' (3).	2 / 3
Completeness	Quotes both the 'what' ('Data structure for annotated matrices in single-cell analysis') and an explicit 'when' ('Use when working with .h5ad files or integrating with the scverse ecosystem'), clearly answering both with an explicit trigger clause.	3 / 3
Trigger Term Quality	Quotes '.h5ad files', 'scverse ecosystem', and 'single-cell analysis' — natural terms a user would actually say when they need this skill, giving good coverage per the 'good coverage of natural terms' anchor.	3 / 3
Distinctiveness Conflict Risk	Quotes 'This is the data format skill—for analysis workflows use scanpy; for probabilistic models use scvi-tools; for population-scale queries use cellxgene-census', which carves a clear niche and actively prevents triggering sibling skills.	3 / 3
	Total	11 / 12 Passed

Validation

87%

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 — 14 / 16 Passed

Validation for skill structure

Criteria	Description	Result
metadata_version	'metadata.version' is missing	Warning
frontmatter_unknown_keys	Unknown frontmatter key(s) found; consider removing or moving to metadata	Warning

	Total	14 / 16 Passed

Repository: synthetic-sciences/openscience
Path: backend/cli/skills/biology/anndata/SKILL.md
Commit: 52845c3

Reviewed: about 11 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.