molfeat
Molecular featurization for ML (100+ featurizers). ECFP, MACCS, descriptors, pretrained models (ChemBERTa), convert SMILES to features, for QSAR and molecular ML.
74
70%
Does it follow best practices?
Impact
78%
1.41xAverage score across 3 eval scenarios
Advisory
Suggest reviewing before use
Optimize this skill with Tessl
npx tessl skill review --optimize ./scientific-skills/molfeat/SKILL.mdQuality
Discovery
82%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 strong, domain-specific description with excellent specificity and trigger term coverage for cheminformatics users. Its main weakness is the lack of an explicit 'Use when...' clause, which would help Claude know precisely when to select this skill. The technical terms are appropriate for the target audience and create a very distinct skill profile.
Suggestions
Add an explicit 'Use when...' clause, e.g., 'Use when the user needs to convert molecular structures to numerical features, mentions SMILES strings, fingerprints, molecular descriptors, or cheminformatics featurization for machine learning.'
| Dimension | Reasoning | Score |
|---|---|---|
Specificity | Lists multiple specific concrete actions and tools: ECFP, MACCS, descriptors, pretrained models (ChemBERTa), convert SMILES to features, QSAR, and molecular ML. These are concrete, named capabilities. | 3 / 3 |
Completeness | The 'what' is well covered (molecular featurization with specific methods), and the 'for QSAR and molecular ML' implies when, but there is no explicit 'Use when...' clause with trigger guidance. Per rubric, missing explicit trigger guidance caps completeness at 2. | 2 / 3 |
Trigger Term Quality | Excellent coverage of natural domain-specific terms users would actually use: 'ECFP', 'MACCS', 'descriptors', 'ChemBERTa', 'SMILES', 'featurization', 'QSAR', 'molecular ML', 'featurizers'. These are precisely the terms a cheminformatics user would search for. | 3 / 3 |
Distinctiveness Conflict Risk | Highly distinctive niche — molecular featurization for ML is a very specific domain. The named methods (ECFP, MACCS, ChemBERTa, SMILES) make it extremely unlikely to conflict with other skills. | 3 / 3 |
Total | 11 / 12 Passed |
Implementation
57%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 code examples and good progressive disclosure to reference files, but it is far too verbose for its purpose. It over-explains concepts Claude already understands (ML task types, what fingerprints are for, basic Python patterns), repeats similar patterns across sections, and includes unnecessary framing sections. Trimming the content by 50-60% while preserving the code examples and reference pointers would significantly improve it.
Suggestions
Remove the 'When to Use This Skill' section entirely—Claude can infer applicability from the content itself.
Collapse the 'Core Concepts' section into a brief table or 3-line summary showing Calculator vs Transformer vs PretrainedTransformer with one code snippet each, removing the explanatory prose about when to use each.
Integrate validation steps into workflows: after featurization, check for None values (from failed molecules) and verify feature matrix shape before passing to ML models.
Deduplicate overlapping content between 'Choosing the Right Featurizer', 'Common Featurizers Reference' table, and 'Common Workflows'—consolidate into a single decision guide with the reference table.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The skill is extremely verbose at ~350+ lines. It explains concepts Claude already knows (what QSAR is, what virtual screening is, what similarity searching is), includes a lengthy 'When to Use This Skill' section that's unnecessary, and repeats similar code patterns multiple times. The 'Core Concepts' section explains basic OOP hierarchy that could be shown by example alone. Many sections like 'Choosing the Right Featurizer' and 'Common Workflows' have significant overlap. | 1 / 3 |
Actionability | The skill provides fully executable, copy-paste ready code examples throughout. Every featurizer recommendation comes with concrete instantiation code, and complete workflows (QSAR, virtual screening, similarity search, sklearn pipeline) are provided with real, runnable Python code. | 3 / 3 |
Workflow Clarity | Workflows are presented as code examples but lack explicit validation checkpoints. For instance, the QSAR workflow doesn't validate that featurization succeeded (checking for None values from failed molecules) before training. The virtual screening pipeline doesn't verify feature matrix integrity. Error handling is mentioned separately but not integrated into the workflow steps. | 2 / 3 |
Progressive Disclosure | The skill has a clear overview structure with well-signaled references to three separate files (api_reference.md, available_featurizers.md, examples.md), each with clear descriptions of when to load them. The grep tip for searching references is a nice touch. References are one level deep and clearly organized. | 3 / 3 |
Total | 9 / 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 (510 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.