deepchem
Molecular ML with diverse featurizers and pre-built datasets. Use for property prediction (ADMET, toxicity) with traditional ML or GNNs when you want extensive featurization options and MoleculeNet benchmarks. Best for quick experiments with pre-trained models, diverse molecular representations. For graph-first PyTorch workflows use torchdrug; for benchmark datasets use pytdc.
79
75%
Does it follow best practices?
Impact
83%
1.36xAverage score across 3 eval scenarios
Advisory
Suggest reviewing before use
Optimize this skill with Tessl
npx tessl skill review --optimize ./scientific-skills/deepchem/SKILL.mdQuality
Discovery
100%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 an excellent skill description that clearly communicates specific capabilities (molecular featurization, property prediction, MoleculeNet benchmarks), includes strong domain-specific trigger terms, and explicitly addresses both when to use this skill and when to use alternatives. The cross-referencing of related skills (torchdrug, pytdc) is particularly effective for reducing conflict risk in a multi-skill environment.
| Dimension | Reasoning | Score |
|---|---|---|
Specificity | Lists multiple specific concrete actions and capabilities: molecular featurizers, pre-built datasets, property prediction (ADMET, toxicity), traditional ML, GNNs, MoleculeNet benchmarks, pre-trained models, diverse molecular representations. | 3 / 3 |
Completeness | Clearly answers both 'what' (molecular ML with featurizers, property prediction, pre-built datasets) and 'when' ('Use for property prediction... when you want extensive featurization options and MoleculeNet benchmarks'). Also includes explicit guidance on when NOT to use it, pointing to alternatives. | 3 / 3 |
Trigger Term Quality | Includes strong natural keywords users would say: 'molecular ML', 'ADMET', 'toxicity', 'property prediction', 'featurizers', 'MoleculeNet', 'GNNs', 'molecular representations', 'pre-trained models'. Also differentiates from related tools (torchdrug, pytdc). | 3 / 3 |
Distinctiveness Conflict Risk | Highly distinctive with clear niche (molecular ML featurization and MoleculeNet benchmarks) and explicitly differentiates from competing skills by naming alternatives: 'For graph-first PyTorch workflows use torchdrug; for benchmark datasets use pytdc.' | 3 / 3 |
Total | 12 / 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 is highly actionable with excellent executable code examples and clear model/featurizer selection guidance, but it is far too verbose for a SKILL.md file. Much of the content (detailed API tables, pitfalls, best practices patterns) duplicates what should live in the referenced files. The workflows lack validation checkpoints, and scaffold splitting advice is repeated three separate times.
Suggestions
Reduce the main file to ~100-150 lines by moving the model selection table, featurizer decision tree, common pitfalls, and best practices patterns into the referenced files (api_reference.md, workflows.md).
Remove the 'When to Use This Skill' section entirely—this information belongs in frontmatter metadata, not the body, and Claude doesn't need to be told when molecular ML is relevant.
Add validation checkpoints to workflows, e.g., checking dataset.X.shape after loading, verifying featurization didn't produce NaN values, and checking train/test score gap for overfitting before making predictions.
Consolidate the repeated scaffold splitting advice into a single prominent callout rather than mentioning it in three separate sections.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Extremely verbose at ~400+ lines. Explains concepts Claude already knows (what SMILES are, what random forests are, what overfitting is), includes redundant 'When to Use This Skill' section listing obvious use cases, repeats scaffold splitting advice 3+ times, and provides extensive tables/guides that belong in reference files rather than the main skill. | 1 / 3 |
Actionability | All code examples are concrete, executable, and copy-paste ready with proper imports. Includes complete workflows from data loading through evaluation, CLI examples for scripts, and specific parameter values. The decision trees and selection guides provide clear decision-making frameworks. | 3 / 3 |
Workflow Clarity | Three complete workflows (A, B, C) are clearly sequenced with numbered steps, but none include validation checkpoints or error recovery loops. For ML workflows involving data transformations and model training, there's no guidance on verifying data loaded correctly, checking for featurization failures, or validating model outputs before proceeding. | 2 / 3 |
Progressive Disclosure | References to `references/api_reference.md`, `references/workflows.md`, and `scripts/` directory are well-signaled, but the main file contains far too much inline content that should be in those reference files (full model selection tables, complete featurizer decision trees, extensive best practices, common pitfalls). The overview should be much leaner with more content pushed to references. | 2 / 3 |
Total | 8 / 12 Passed |
Validation
81%Checks the skill against the spec for correct structure and formatting. All validation checks must pass before discovery and implementation can be scored.
Validation — 9 / 11 Passed
Validation for skill structure
| Criteria | Description | Result |
|---|---|---|
skill_md_line_count | SKILL.md is long (596 lines); consider splitting into references/ and linking | Warning |
metadata_version | 'metadata.version' is missing | Warning |
Total | 9 / 11 Passed | |
- Repository
- K-Dense-AI/claude-scientific-skills
- Commit
b58ad7e
- 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.