dstack-prototyping

Use with the dstack skill for model-serving work when the image, serving command, resources, backend/fleet choice, or service behavior is not proven. Guides task-first prototyping on real hardware, choosing fleets/backends that can reuse idle instances and caches, checking vLLM/SGLang sources, and verifying the final dstack service with a model request.

Quality

88%

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

85%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 a tight, actionable workflow with explicit validation and feedback loops and clean delegation of syntax details to the dstack skill and external docs. Minor conciseness tightening and a fully-complete example serving command would push it to the top anchor.

Suggestions

Tighten the 'Choose Where To Run' opening paragraph and drop redundant asides like 'or an equivalent idle command' to reduce tokens without losing clarity.

Provide one fully concrete (non-`...`) example serving command or point to the specific vLLM/SGLang recipe that supplies the real flags, so the actionability example is copy-paste ready.

Consider moving the detailed PD-disaggregation specifics into a short reference section or file so the main flow stays an overview.

Dimension	Reasoning	Score
Conciseness	The body is lean and procedural with concrete operational specifics, but a few explanatory asides (e.g. 'or an equivalent idle command', the verbose opening of 'Choose Where To Run') could be trimmed without losing clarity.	4 / 5
Actionability	Concrete, executable guidance is present (`sleep infinity`, the nohup redirect command, `dstack fleet get <fleet name> --json`, `placement: cluster`), but the example serving command uses `...` and some steps remain guidance rather than literal copy-paste commands.	4 / 5
Workflow Clarity	A clear task→verify→service→verify sequence is given with explicit validation (require a real model request, not just GPU visibility/logs) and an explicit feedback loop (return to a task if service verification fails), plus a verification checklist of config dimensions.	5 / 5
Progressive Disclosure	The body is well-sectioned, delegates CLI/YAML syntax to the /dstack skill, and signals one-level-deep external references (dstack.ai docs, vLLM/SGLang sources) clearly, with no bundle files needed and easy navigation.	5 / 5
	Total	18 / 20 Passed

Description

92%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 specific, complete, and well-scoped, clearly stating both the capability and the triggering conditions while carving out a distinct sub-niche of dstack usage. Trigger term coverage is strong but slightly specialized rather than comprehensive.

Dimension	Reasoning	Score
Specificity	Lists multiple concrete actions — 'task-first prototyping on real hardware', 'choosing fleets/backends that can reuse idle instances and caches', 'checking vLLM/SGLang sources', and 'verifying the final dstack service with a model request' — giving comprehensive coverage of the workflow.	5 / 5
Completeness	Explicitly answers both what (prototyping, fleet/backend selection, source checking, service verification) and when ('Use with the dstack skill for model-serving work when the image, serving command, resources, backend/fleet choice, or service behavior is not proven') with concrete trigger conditions.	5 / 5
Trigger Term Quality	Good coverage of domain-natural terms ('model-serving work', 'serving command', 'backend/fleet choice', 'vLLM/SGLang'), though a few synonyms and variations a user might naturally say are missing, keeping it just below the comprehensive anchor.	4 / 5
Distinctiveness Conflict Risk	Scoped to a precise niche — dstack model-serving with unproven configuration — and explicitly defers CLI/syntax to the dstack skill, giving it clear distinct triggers and minimal conflict risk.	5 / 5
	Total	19 / 20 Passed

Validation

100%

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 — 16 / 16 Passed

Validation for skill structure

No warnings or errors.

Repository: dstackai/dstack
Path: skills/dstack-prototyping/SKILL.md
Commit: 87e2fd6

Reviewed: 3 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.