serverless-modal
Run GPU workloads on Modal — training, fine-tuning, inference, batch processing. Zero-config serverless: no SSH, no Docker, auto scale-to-zero. Use when user says "modal run", "modal training", "modal inference", "deploy to modal", "need a GPU", "run on modal", "serverless GPU", or needs remote GPU compute.
71
88%
Does it follow best practices?
Impact
—
No eval scenarios have been run
Advisory
Suggest reviewing before use
Quality
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 hits all the marks. It provides specific capabilities (training, fine-tuning, inference, batch processing), differentiating characteristics (zero-config, no SSH/Docker, scale-to-zero), and a comprehensive 'Use when' clause with natural trigger terms. The description is concise yet thorough, and clearly distinguishable from other skills.
| Dimension | Reasoning | Score |
|---|---|---|
Specificity | Lists multiple specific concrete actions: training, fine-tuning, inference, batch processing. Also specifies key characteristics like 'zero-config serverless', 'no SSH, no Docker, auto scale-to-zero'. | 3 / 3 |
Completeness | Clearly answers both 'what' (run GPU workloads on Modal — training, fine-tuning, inference, batch processing with zero-config serverless) and 'when' (explicit 'Use when' clause with multiple specific trigger phrases). | 3 / 3 |
Trigger Term Quality | Excellent coverage of natural trigger terms users would say: 'modal run', 'modal training', 'modal inference', 'deploy to modal', 'need a GPU', 'run on modal', 'serverless GPU', 'remote GPU compute'. These are highly natural phrases a user would actually type. | 3 / 3 |
Distinctiveness Conflict Risk | Highly distinctive — 'Modal' is a specific platform, and the triggers are tightly scoped to Modal-related commands and GPU compute needs. Unlikely to conflict with generic coding or deployment skills. | 3 / 3 |
Total | 12 / 12 Passed |
Implementation
77%Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
This is a strong, highly actionable skill with excellent executable code patterns covering the major Modal use cases. The workflow is well-sequenced with cost estimation as a mandatory checkpoint. The main weakness is length — at ~300 lines with no bundle files, it could benefit from splitting detailed patterns and pricing into separate reference files, and trimming some redundant explanations (e.g., duplicate security warnings, overview comparisons to other platforms).
Suggestions
Remove or significantly trim the Overview section's comparison to other platforms (vast.ai, Lightning) — Claude can infer trade-offs; focus on Modal-specific setup facts only.
Extract the six code patterns (A-F) into a separate PATTERNS.md reference file, keeping only Pattern A inline as the primary example with links to the rest.
Consolidate the duplicate cost protection / security warnings into a single prominent callout rather than repeating at top and bottom.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The skill is fairly comprehensive but includes some unnecessary verbosity — the overview section explains what serverless means and compares to other platforms (Claude knows this), the pricing table is extensive, and there's some redundancy (cost protection advice appears twice, security warning repeated). However, most content earns its place as Modal-specific knowledge Claude wouldn't have. | 2 / 3 |
Actionability | Excellent actionability with fully executable, copy-paste-ready code for six distinct patterns (one-shot, web API, vLLM, batch, LoRA, multi-GPU). CLI commands are concrete, authentication steps are specific, and the cost estimation template provides a clear fill-in-the-blank format. Every pattern includes runnable Python code. | 3 / 3 |
Workflow Clarity | Clear 6-step workflow (Analyze→Generate→Run→Verify→Collect→Cleanup) with explicit validation via `modal app logs` and `modal app list`. The cost estimation checkpoint before every run acts as a validation gate. Cleanup notes that auto-scale-to-zero handles most cases but provides explicit commands for volumes and deployed services. | 3 / 3 |
Progressive Disclosure | The skill is quite long (~300 lines) with all content inline — the pricing table, six code patterns, CLI reference, CLAUDE.md example, and composability section could benefit from being split into referenced files. However, the content is well-organized with clear headers and the external documentation links at the bottom are helpful. No bundle files exist to offload content to. | 2 / 3 |
Total | 10 / 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
66b974e
- 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.