ml-model-training

Train ML models with scikit-learn, PyTorch, TensorFlow. Use for classification/regression, neural networks, hyperparameter tuning, or encountering overfitting, underfitting, convergence issues.

Quality

77%

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

Fix and improve this skill with Tessl

tessl review fix ./plugins/ml-model-training/skills/ml-model-training/SKILL.md

Quality

Content

72%

Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.

A highly actionable, well-structured skill body with executable code and clean progressive disclosure to two real reference files. Its main weaknesses are duplicated scaling code, restated well-known ML concepts, and a workflow that lists steps without explicit validation checkpoints.

Suggestions

Remove the duplicated StandardScaler block in 'Known Issues Prevention #1' by referencing the Data Preparation snippet instead, to improve token efficiency.

Add explicit validation checkpoints to the training workflow (e.g., 'After data prep: verify no leakage; After training: confirm val loss is decreasing before tuning') to lift workflow clarity.

Trim the 'Problem' explanations of well-known concepts (data leakage, overfitting, reproducibility) and keep only the corrective code, since Claude already knows these.

Define the missing tensor variables in the PyTorch loop (show the torch.from_numpy / DataLoader conversion) so the example is fully self-contained.

Dimension	Reasoning	Score
Conciseness	Mostly efficient and code-first, but the StandardScaler block is duplicated (Data Preparation and Known Issues #1) and the Problem/Solution framing restates well-known concepts (data leakage, overfitting) Claude already knows; not 1 because there is no padded prose intro, not 3 due to the redundancy and concept re-explanation.	2 / 3
Actionability	Substantially executable, copy-paste-ready code for data prep, sklearn and PyTorch training, class weighting, seeds, and GridSearchCV; the only gap is the undefined 'X_train_tensor'/'y_train_tensor' in the PyTorch loop, which is minor relative to the bulk.	3 / 3
Workflow Clarity	A clear 5-step sequence is listed ('Data Preparation -> ... -> Evaluation'), but validation checkpoints between steps are implicit (no 'check for leakage' or 'confirm val metric' gates), matching the listed-steps-with-validation-gaps anchor.	2 / 3
Progressive Disclosure	The body is an overview with well-signaled, one-level-deep references to two real files (pytorch-training.md, tensorflow-keras.md), each annotated with its contents, plus a 'When to Load References' navigation section, matching the clear-overview anchor.	3 / 3
	Total	10 / 12 Passed

Description

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.

A strong, trigger-rich description that clearly states what the skill does and when to use it, with natural user-facing keywords. It is held back slightly by a single broad action verb and some overlap risk with neighboring ML skills.

Suggestions

Lead with multiple concrete action verbs to lift specificity, e.g. 'Train, tune, and evaluate ML models with scikit-learn, PyTorch, TensorFlow' instead of the single verb 'Train'.

Tighten distinctiveness by scoping the trigger away from adjacent skills, e.g. 'Use when training or fine-tuning models (not for data cleaning or deployment alone)...'.

Consider adding a couple of natural phrasings users say ('model won't converge', 'model is overfitting') to broaden trigger coverage further.

Dimension	Reasoning	Score
Specificity	Names the domain and three concrete frameworks ('scikit-learn, PyTorch, TensorFlow'), but the action set is a single verb ('Train ML models') rather than the multiple distinct action verbs of the score-3 anchor; it is concrete rather than vague, so above 1.	2 / 3
Completeness	Explicitly answers what ('Train ML models with scikit-learn, PyTorch, TensorFlow') and when ('Use for ...'), with an explicit trigger clause, satisfying the both-what-and-when anchor.	3 / 3
Trigger Term Quality	Strong natural-term coverage users would actually say: 'classification/regression, neural networks, hyperparameter tuning, or encountering overfitting, underfitting, convergence issues', matching the good-coverage anchor.	3 / 3
Distinctiveness Conflict Risk	It targets a clear ML-training niche with specific framework and issue triggers, but the broad 'Train ML models' phrasing could overlap adjacent skills (data preprocessing, model evaluation, deployment), so it does not fully meet the 'unlikely to conflict' bar of 3.	2 / 3
	Total	10 / 12 Passed

Validation

93%

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 — 15 / 16 Passed

Validation for skill structure

Criteria	Description	Result
metadata_version	'metadata.version' is missing	Warning

	Total	15 / 16 Passed

Repository: secondsky/claude-skills
Path: plugins/ml-model-training/skills/ml-model-training/SKILL.md
Commit: c4889f6

Reviewed: about 2 hours 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.