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

82%

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%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 well-structured, highly actionable skill body with excellent progressive disclosure and executable examples. It is held back by some redundant conceptual padding, undeprecated time-sensitive version info, and a training workflow missing validation checkpoints.

Suggestions

Move the 'Theoretical Foundations' explanation of variational inference/VAEs into references/theoretical-foundations.md and keep only a one-line pointer, since Claude already knows these concepts.

Add a validation checkpoint to the training workflow (e.g., inspect training loss for convergence or check latent representation shape/quality before downstream analysis).

Relocate version-specific details (1.4.3, May 2026, '1.4 dropped older versions') into a dedicated versioning/compatibility note so time-sensitive information does not pad the main body.

Dimension	Reasoning	Score
Conciseness	Mostly efficient with executable code, but the 'Theoretical Foundations' section explains variational inference/VAEs Claude already knows, 'Key Design Principles' repeats the raw-counts point, and time-sensitive version info (1.4.3, May 2026) sits outside any deprecated/old-patterns section.	3 / 5
Actionability	Provides several complete, copy-paste-ready code blocks covering the core workflow, differential expression, model save/load, and batch correction — concrete and executable across the common cases.	5 / 5
Workflow Clarity	A clear numbered 6-step training workflow is present, but it lacks validation/convergence checkpoints for what is a batch training operation; no feedback loop for inspecting loss or verifying the learned representation.	3 / 5
Progressive Disclosure	SKILL.md is a concise overview that signals one-level-deep references ('See references/… for:') for each modality, and every referenced file exists in references/ with content appropriately split by topic.	5 / 5
	Total	16 / 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, well-scoped description that answers both what and when with concrete actions and explicit de-confliction guidance against scanpy. Minor keyword-synonym gaps keep trigger quality just short of perfect.

Dimension	Reasoning	Score
Specificity	Lists multiple concrete actions — probabilistic batch correction, transfer learning, differential expression with uncertainty, and multi-modal integration — giving comprehensive coverage of the framework's capabilities.	5 / 5
Completeness	Explicitly states what the skill does ('Deep generative models for single-cell omics') and when to use it ('Use when you need…', 'Best for…'), with concrete trigger phrases and a negative boundary.	5 / 5
Trigger Term Quality	Includes natural terms users would say ('batch correction', 'batch effects', 'differential expression', 'integration', 'multimodal data') but omits some common synonyms and modality-specific triggers like 'single-cell RNA-seq' or 'ATAC-seq'.	4 / 5
Distinctiveness Conflict Risk	Occupies a clear niche (probabilistic single-cell modeling) and actively reduces conflict risk via 'For standard analysis pipelines use scanpy', steering generic tasks away.	5 / 5
	Total	19 / 20 Passed

Validation

100%

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 — 16 / 16 Passed

Validation for skill structure

No warnings or errors.

Repository: K-Dense-AI/scientific-agent-skills
Path: skills/scvi-tools/SKILL.md
Commit: b2a92ba

Reviewed: 3 days 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.