scvi-tools

Deep generative models for single-cell omics. Use when you need probabilistic batch correction (scVI), transfer learning, differential expression with uncertainty, or multi-modal integration (TOTALVI, MultiVI). Best for advanced modeling, batch effects, multimodal data. For standard analysis pipelines use scanpy.

Quality

86%

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

80%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.

An actionable, well-structured skill with executable examples and clean progressive disclosure to verified reference files. The main gap is workflow clarity: the model-training and batch-correction workflow lacks explicit validation/verification checkpoints.

Suggestions

Add a validation checkpoint after training (e.g., inspect `model.history`/reconstruction loss or check that the latent representation separates expected groups) before downstream UMAP/leiden.

Trim the Theoretical Foundations bullet list or fold it entirely into the reference, since VAE/variational-inference concepts are already known to Claude.

Add an explicit convergence/sanity-check step in the batch-correction example rather than implying the latent is corrected by comment alone.

Dimension	Reasoning	Score
Conciseness	Mostly lean with compact, executable code and brief overviews, but the Theoretical Foundations section re-explains VAEs/variational inference and the batch-correction task partially re-treads the main workflow; minor trimming would lift it to a 5.	4 / 5
Actionability	Provides copy-paste-ready, executable Python for the full setup→train→extract workflow plus differential expression, save/load, and batch correction — concrete examples covering the common cases.	5 / 5
Workflow Clarity	The Typical Workflow is a clearly numbered sequence (1–6) with code, but it has no validation checkpoints (e.g., confirm training converged, verify the latent representation) for what are effectively batch/model-training operations; missing validation caps this at 3.	3 / 5
Progressive Disclosure	SKILL.md is a clear overview with well-signaled, one-level-deep references ("See `references/models-scrna-seq.md` for:" etc.); all eight referenced files exist and model details are appropriately split out, making navigation easy.	5 / 5
	Total	17 / 20 Passed

Description

92%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, specific description that answers what and when with concrete model-named triggers and an explicit boundary against the sibling scanpy skill. Trigger-term coverage is good but could add common synonyms and file extensions to reach the top anchor.

Dimension	Reasoning	Score
Specificity	Names the domain ("single-cell omics") and multiple concrete capabilities with model names — "probabilistic batch correction (scVI)", "transfer learning", "differential expression with uncertainty", "multi-modal integration (TOTALVI, MultiVI)" — giving comprehensive coverage of specific actions.	5 / 5
Completeness	Explicitly answers both what ("Deep generative models for single-cell omics") and when ("Use when you need probabilistic batch correction...") with concrete trigger phrases, plus an explicit negative boundary ("For standard analysis pipelines use scanpy").	5 / 5
Trigger Term Quality	Good coverage of natural terms users would say ("batch correction", "transfer learning", "differential expression", "multi-modal integration", "batch effects"), but it omits common synonyms/extensions (e.g., "scRNA-seq", "CITE-seq", "AnnData", ".h5ad") that anchor 5 expects.	4 / 5
Distinctiveness Conflict Risk	Occupies a clear niche (probabilistic deep generative models for single-cell omics) with distinct triggers, and the explicit hand-off to scanpy for standard pipelines minimizes overlap/conflict risk.	5 / 5
	Total	19 / 20 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/scvi-tools/SKILL.md
Commit: 3a6c3a9

Reviewed: 1 day 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.