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.
76
70%
Does it follow best practices?
Impact
81%
1.09xAverage score across 3 eval scenarios
Advisory
Suggest reviewing before use
Optimize this skill with Tessl
npx tessl skill review --optimize ./scientific-skills/anndata/SKILL.mdQuality
Discovery
89%Based on the skill's description, can an agent find and select it at the right time? Clear, specific descriptions lead to better discovery.
This is a well-crafted description with excellent trigger terms, clear 'when' guidance, and outstanding disambiguation from related skills. Its main weakness is the lack of specific concrete actions—it describes what the skill is about rather than listing specific operations it enables (e.g., reading, writing, subsetting, or converting AnnData objects).
Suggestions
Add specific concrete actions the skill covers, e.g., 'Read, write, subset, and manipulate AnnData objects for annotated matrices in single-cell analysis.'
| Dimension | Reasoning | Score |
|---|---|---|
Specificity | It names the domain ('annotated matrices in single-cell analysis') and mentions the data format, but does not list specific concrete actions like 'read', 'write', 'convert', 'subset', or 'manipulate AnnData objects'. The description is more about what the skill represents than what it does. | 2 / 3 |
Completeness | Clearly answers both 'what' (data structure for annotated matrices in single-cell analysis) and 'when' ('Use when working with .h5ad files or integrating with the scverse ecosystem'). It also explicitly disambiguates from related skills, which adds to completeness. | 3 / 3 |
Trigger Term Quality | Includes strong natural trigger terms: '.h5ad files', 'scverse ecosystem', 'annotated matrices', 'single-cell analysis', and cross-references related tools (scanpy, scvi-tools, cellxgene-census) which helps with disambiguation. These are terms users working in this domain would naturally use. | 3 / 3 |
Distinctiveness Conflict Risk | Excellent distinctiveness—explicitly differentiates itself from scanpy (analysis workflows), scvi-tools (probabilistic models), and cellxgene-census (population-scale queries). The '.h5ad files' and 'data format skill' framing create a clear, non-overlapping niche. | 3 / 3 |
Total | 11 / 12 Passed |
Implementation
50%Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
The skill provides excellent actionable code examples covering the full breadth of AnnData usage, but is far too verbose for a skill file. It duplicates content between Quick Start, Core Capabilities, and Common Workflows sections, includes Scanpy-heavy workflows that the description explicitly says belong elsewhere, and explains concepts Claude already understands. The reference file structure is good but undermined by excessive inline content.
Suggestions
Cut the content by ~60%: remove the 'When to Use This Skill' section, the 'Overview' paragraph, the Scanpy/Muon integration examples (per the description, those belong in their own skills), and deduplicate between Quick Start and Core Capabilities sections.
Remove the Common Workflows section entirely—the scRNA-seq and batch integration workflows are Scanpy workflows, not AnnData data-structure skills, and the description explicitly defers those to scanpy.
Add validation checkpoints to remaining workflows, e.g., after reading data verify shape/contents with `print(adata)`, after concatenation verify batch labels are correct, after filtering verify expected cell counts.
Consolidate the 'common commands' snippets in Core Capabilities with the Quick Start examples—currently the same patterns (read_h5ad, write_h5ad, subsetting) appear in multiple places.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The skill is extremely verbose at ~300+ lines. It explains concepts Claude already knows (what AnnData is, what single-cell genomics is), includes a full 'When to Use This Skill' section that restates the description, provides lengthy Scanpy workflows that belong in a separate scanpy skill (as the description itself notes), and repeats common commands in both Quick Start and Core Capabilities sections. | 1 / 3 |
Actionability | The skill provides fully executable, copy-paste ready code examples throughout—creating AnnData objects, reading/writing files, subsetting, concatenation, batch integration, and troubleshooting patterns. All code is concrete with real imports and realistic parameters. | 3 / 3 |
Workflow Clarity | Multi-step workflows like the scRNA-seq analysis and batch integration are clearly sequenced with numbered steps, but they lack validation checkpoints—there's no verification that data loaded correctly, no checks after filtering steps, and no error recovery guidance within the workflows themselves. | 2 / 3 |
Progressive Disclosure | The skill references separate files (references/data_structure.md, references/io_operations.md, etc.) which is good progressive disclosure, but the main file itself contains too much inline content that duplicates what those reference files likely cover. The Quick Start, Core Capabilities common commands, Common Workflows, and Integration sections create significant redundancy. | 2 / 3 |
Total | 8 / 12 Passed |
Validation
90%Checks the skill against the spec for correct structure and formatting. All validation checks must pass before discovery and implementation can be scored.
Validation — 10 / 11 Passed
Validation for skill structure
| Criteria | Description | Result |
|---|---|---|
metadata_version | 'metadata.version' is missing | Warning |
Total | 10 / 11 Passed | |
- Repository
- K-Dense-AI/claude-scientific-skills
- Commit
25e1c0f
- 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.