model-checkpoint-manager
Model Checkpoint Manager - Auto-activating skill for ML Training. Triggers on: model checkpoint manager, model checkpoint manager Part of the ML Training skill category.
36
3%
Does it follow best practices?
Impact
96%
1.00xAverage score across 3 eval scenarios
Passed
No known issues
Optimize this skill with Tessl
npx tessl skill review --optimize ./planned-skills/generated/07-ml-training/model-checkpoint-manager/SKILL.mdQuality
Discovery
7%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 description is extremely weak—it is essentially just a title repeated as a trigger term with no substantive content. It fails to describe any concrete capabilities, lacks natural trigger terms users would actually say, and provides no explicit guidance on when Claude should select this skill. The description reads as auto-generated boilerplate rather than a useful skill selector.
Suggestions
Add specific concrete actions the skill performs, e.g., 'Saves, loads, compares, and manages ML model checkpoints during training. Supports resuming interrupted training runs and selecting best-performing model snapshots.'
Add an explicit 'Use when...' clause with natural trigger terms, e.g., 'Use when the user mentions checkpoints, saving model weights, resuming training, .ckpt or .pt files, epoch snapshots, or model selection.'
Remove the duplicate trigger term ('model checkpoint manager' is listed twice) and replace with diverse natural language variations users would actually use, such as 'save checkpoint', 'load weights', 'best model', 'training resume'.
| Dimension | Reasoning | Score |
|---|---|---|
Specificity | The description names a domain ('ML Training') and a concept ('Model Checkpoint Manager') but does not describe any concrete actions. There are no verbs indicating what the skill actually does—no mention of saving, loading, comparing, deleting, or managing checkpoints. | 1 / 3 |
Completeness | The description fails to answer 'what does this do' beyond the name itself, and the 'when' clause is just a self-referential trigger ('model checkpoint manager'). There is no explicit 'Use when...' guidance with meaningful triggers. | 1 / 3 |
Trigger Term Quality | The only trigger terms listed are 'model checkpoint manager' repeated twice. There are no natural user keywords like 'checkpoint', 'save model', 'resume training', 'model weights', '.ckpt', '.pt', 'epoch', or other variations a user would naturally say. | 1 / 3 |
Distinctiveness Conflict Risk | The term 'Model Checkpoint Manager' is somewhat specific to a niche (ML checkpoint management), which provides some distinctiveness. However, the lack of concrete actions and the vague 'ML Training' category could cause overlap with other ML-related skills. | 2 / 3 |
Total | 5 / 12 Passed |
Implementation
0%Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
This skill is an empty template with no substantive content. It repeatedly references 'model checkpoint manager' without ever explaining what it does, how to implement it, or providing any code examples. It fails on every dimension because it contains zero actionable information about checkpoint management in ML training.
Suggestions
Add concrete, executable code examples for saving and loading model checkpoints in PyTorch and/or TensorFlow (e.g., `torch.save(model.state_dict(), 'checkpoint.pt')`).
Define a clear workflow for checkpoint management: when to save, naming conventions, how to resume training from a checkpoint, and validation steps to verify checkpoint integrity.
Include specific guidance on best practices such as saving optimizer state, tracking metrics alongside checkpoints, implementing early stopping, and managing disk space with checkpoint rotation.
Remove all boilerplate sections (Purpose, When to Use, Example Triggers) that contain no real information and replace with actionable content.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The content is entirely filler and boilerplate. It explains nothing Claude doesn't already know, repeats 'model checkpoint manager' excessively, and provides zero substantive information about how to actually manage model checkpoints. | 1 / 3 |
Actionability | There is no concrete code, no specific commands, no examples of checkpoint saving/loading, no library references, and no executable guidance whatsoever. Every section is vague and abstract. | 1 / 3 |
Workflow Clarity | No workflow steps are defined. Claims to provide 'step-by-step guidance' but includes none. There are no sequences, no validation checkpoints, and no actual process described for managing model checkpoints. | 1 / 3 |
Progressive Disclosure | The content is a monolithic block of generic placeholder text with no references to detailed materials, no links to related files, and no meaningful structure beyond boilerplate headings that contain no real content. | 1 / 3 |
Total | 4 / 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 |
|---|---|---|
allowed_tools_field | 'allowed-tools' contains unusual tool name(s) | Warning |
frontmatter_unknown_keys | Unknown frontmatter key(s) found; consider removing or moving to metadata | Warning |
Total | 9 / 11 Passed | |
- Commit
3076d78
- 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.