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.

Quality

83%

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

Quality

Content

71%Weight 40%Scale 1-5

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

The content is highly actionable and information-dense, with a clear workflow and strong cost-safety checkpoints, but it is a monolithic single-file skill: pricing tables, CLI reference, and benchmark data are inlined rather than split into one-level-deep reference files, and the cost-protection warning is duplicated.

Suggestions

Move the pricing table, benchmark cost comparison, and CLI reference into references/ files (e.g. references/pricing.md, references/cli.md) and link them from a lean overview section — this would cut the body substantially and fix progressive disclosure.

Deduplicate the card-binding/spending-limit guidance: keep one canonical security warning instead of repeating it in both the Authentication section and the closing 'Cost protection' blockquote.

Make Patterns D and E self-contained (or explicitly state they reuse the `image`/`volume` definitions from Pattern A) so every code example is copy-paste executable, and add a short error-recovery note (e.g. check `modal app logs` on failure and reduce timeout/GPU tier).

Dimension	Reasoning	Score
Conciseness	The body is dense with tables, code, and terse bullets and explains nothing Claude already knows, but the card-binding/spending-limit warning appears twice (Authentication blockquote and closing 'Cost protection' blockquote) and the benchmark-comparison table could be trimmed. Efficient with minor duplication — anchor 4, not 5.	4 / 5
Actionability	Patterns A–C are complete, executable launchers with exact run/deploy commands, and the CLI reference is copy-paste ready. However, Patterns D and E reference `image`/`volume` variables never defined in those snippets, leaving minor gaps — anchor 4, not 5.	4 / 5
Workflow Clarity	A clear 6-step sequence (analyze/estimate → generate → run → verify → collect → cleanup) with a required pre-run cost-estimate checkpoint and a verify/monitor step. There is no explicit error-recovery loop (what to inspect when a run fails), so it falls short of the anchor-5 feedback-loop pattern but is well above anchor 3.	4 / 5
Progressive Disclosure	No bundle files exist (no references/, scripts/, or assets/), and the ~336-line monolithic body inlines content that belongs in separate reference files — the full pricing table, benchmark cost comparison, and CLI reference. Section headers are well-organized, but content that should be split out is inline, matching anchor 3.	3 / 5
	Total	15 / 20 Passed

Description

96%Weight 40%Scale 1-5

Based on the skill's description, can an agent find and select it at the right time? Clear, specific descriptions lead to better discovery.

The description is exemplary: concrete actions, comprehensive natural trigger phrases, and an explicit 'Use when' clause. The only weakness is the broad 'need a GPU' / 'serverless GPU' triggers, which risk overlap with other GPU-provider skills in a multi-skill setup.

Dimension	Reasoning	Score
Specificity	"Run GPU workloads on Modal — training, fine-tuning, inference, batch processing" lists multiple concrete actions with zero vague filler, and the third-person imperative voice is used correctly. Coverage of workload types is comprehensive, matching the anchor-5 example structure rather than the 'minor gaps' of anchor 4.	5 / 5
Completeness	It explicitly answers both what ("Run GPU workloads on Modal — training, fine-tuning, inference, batch processing. Zero-config serverless...") and when ("Use when user says...or needs remote GPU compute") with concrete trigger phrases, matching the anchor-5 example.	5 / 5
Trigger Term Quality	Trigger phrases "modal run", "modal training", "modal inference", "deploy to modal", "need a GPU", "run on modal", "serverless GPU", "remote GPU compute" cover the natural phrasings a user would actually say, including both CLI-command forms and plain-language synonyms.	5 / 5
Distinctiveness Conflict Risk	Modal-specific terms ("modal run", "deploy to modal") establish a clear niche, but the generic triggers "need a GPU" and "serverless GPU" could fire for competing GPU-provider skills (the body itself compares vast.ai/Lightning). Mostly distinct with minor overlap risk — anchor 4, not 5.	4 / 5
	Total	19 / 20 Passed

Validation

87%

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 — 14 / 16 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	14 / 16 Passed

Repository: wanshuiyin/Auto-claude-code-research-in-sleep
Path: skills/serverless-modal/SKILL.md
Commit: 2132036

Reviewed: 4 days 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.