skypilot-multi-cloud-orchestration

Multi-cloud orchestration for ML workloads with automatic cost optimization. Use when you need to run training or batch jobs across multiple clouds, leverage spot instances with auto-recovery, or optimize GPU costs across providers.

Quality

81%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

Securityby

Low

Low-risk findings worth noting

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 reference with strong progressive disclosure via real bundle files. It loses points for redundancy that could be tightened and for missing validation checkpoints around destructive and batch operations.

Suggestions

Consolidate the duplicated 'any_of' cloud-fallback and checkpointing examples into single canonical snippets and cross-reference them instead of repeating.

Add an explicit validation/confirmation step before destructive operations (e.g. run 'sky status' to confirm the target before 'sky down', or a dry-run check before the batch sweep loop).

Tighten the 'When to use SkyPilot' / 'Key features' / 'Use alternatives' triad, which overlaps content already demonstrated in the code sections.

Dimension	Reasoning	Score
Conciseness	It assumes Claude's competence (no basic concept explanations) and is concrete throughout, but carries notable redundancy — the 'any_of' cloud fallback appears twice, checkpointing is duplicated between 'Managed jobs' and 'Workflow 1', and spot-instance config recurs — so it 'could be tightened'.	2 / 3
Actionability	Nearly every section ships fully executable YAML and bash commands (e.g. 'sky launch -c hello hello.yaml', the torchrun distributed-training block), making it copy-paste ready.	3 / 3
Workflow Clarity	It reads as a reference catalogue rather than sequenced workflows, and destructive/batch operations like 'sky down' and the 10-iteration hyperparameter sweep loop lack explicit validation or verification checkpoints, capping this dimension at 2.	2 / 3
Progressive Disclosure	The body is a concise overview pointing to two real, one-level-deep references clearly signaled via markdown links ('[Advanced Usage](references/advanced-usage.md)' and '[Troubleshooting](references/troubleshooting.md)').	3 / 3
	Total	10 / 12 Passed

Description

90%

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 description that concretely states capabilities and gives explicit usage triggers with natural keywords. The only ding is the second-person 'you' voice, which costs one specificity point per the rubric guidelines.

Dimension	Reasoning	Score
Specificity	It lists multiple concrete actions ('run training or batch jobs across multiple clouds, leverage spot instances with auto-recovery, or optimize GPU costs'), which would be a 3, but the second-person phrasing 'Use when you need to...' is penalized by one level per the guidelines.	2 / 3
Completeness	It explicitly states what it does ('Multi-cloud orchestration for ML workloads with automatic cost optimization') and when to use it ('Use when you need to run training or batch jobs...').	3 / 3
Trigger Term Quality	It covers natural terms users would actually say — 'training or batch jobs across multiple clouds', 'spot instances with auto-recovery', and 'optimize GPU costs across providers'.	3 / 3
Distinctiveness Conflict Risk	The multi-cloud ML orchestration niche with SkyPilot-specific triggers is clearly distinguishable and unlikely to fire for unrelated skills.	3 / 3
	Total	11 / 12 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
skill_md_line_count	SKILL.md is long (510 lines); consider splitting into references/ and linking	Warning
frontmatter_unknown_keys	Unknown frontmatter key(s) found; consider removing or moving to metadata	Warning

	Total	14 / 16 Passed

Repository: Orchestra-Research/AI-Research-SKILLs
Path: 09-infrastructure/skypilot/SKILL.md
Commit: 773a529

Reviewed: about 12 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.