run-experiment

Deploy and run ML experiments on local or remote GPU servers. Use when user says "run experiment", "deploy to server", "跑实验", or needs to launch training jobs.

Quality

—

Does it follow best practices?

Impact

—

No eval scenarios have been run

Securityby

Risky

Do not use without reviewing

Quality

Content

77%

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

A thorough, actionable multi-environment deployment workflow with strong sequencing and validation. It is let down by a monolithic structure with a dangling external reference and some dense, tighten-able prose.

Suggestions

Move the W&B integration detail and the AGENTS.md example into separate reference files (e.g. references/wandb.md, references/agents-example.md) and link to them, reducing the inline wall of text.

Resolve or remove the dangling ../shared-references/compute-env-contract.md reference, or create the file so the one-level-deep pointer actually resolves.

Tighten the "Environment contract" paragraph by replacing opaque jargon ("seeded kernel witness", "agent-follows-doc pass") with a concise, concrete checklist or a link to an explanation.

Dimension	Reasoning	Score
Conciseness	Most of the body is concrete commands that earn their place, but the dense "Environment contract" paragraph (seeded kernel witness, agent-follows-doc pass) is jargon-heavy and the W&B/AGENTS.md sections are lengthy and could be tightened.	2 / 3
Actionability	It provides copy-paste-ready, executable commands for every environment (ssh nvidia-smi, rsync, git push/pull, screen+conda launch, modal run, local CUDA/MPS) with clearly marked placeholders.	3 / 3
Workflow Clarity	Steps 1–7 are clearly sequenced with explicit validation checkpoints (pre-flight GPU check, verify-launch screen -ls, wandb login check) and feedback loops for destructive ops (auto-destroy only after artifacts persist; do not relaunch blindly on failure).	3 / 3
Progressive Disclosure	Content is well-sectioned but monolithic (~230 lines all inline), and the one external reference to ../shared-references/compute-env-contract.md is a dangling path with no bundle files present; the W&B detail and AGENTS.md example could live in separate referenced files.	2 / 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 concise, third-person description that clearly answers both what the skill does and when to invoke it, with natural multilingual trigger terms. The only weakness is modest action enumeration in the capability statement.

Dimension	Reasoning	Score
Specificity	"Deploy and run ML experiments on local or remote GPU servers" names the domain (ML experiments, GPU servers, local/remote) and concrete actions (deploy, run), but enumerates only two verbs rather than a comprehensive list of the skill's capabilities.	2 / 3
Completeness	It clearly states what it does (deploy and run ML experiments on local/remote GPU servers) and when to use it via an explicit "Use when..." clause with concrete triggers, satisfying both what and when.	3 / 3
Trigger Term Quality	"Use when user says \"run experiment\", \"deploy to server\", \"跑实验\", or needs to launch training jobs" gives good, natural coverage including multilingual (Chinese) phrasings a user would actually say.	3 / 3
Distinctiveness Conflict Risk	The niche (ML experiments on GPU servers, including local/remote) with distinct triggers is specific enough to be unlikely to fire for unrelated skills.	3 / 3
	Total	11 / 12 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: wanshuiyin/Auto-claude-code-research-in-sleep
Commit: fe5963c

Reviewed: about 17 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.