CtrlK
BlogDocsLog inGet started
Tessl Logo

aris-run-experiment

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

63

Quality

75%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

SecuritybySnyk

Passed

No findings from the security scan

Fix and improve this skill with Tessl

tessl review fix ./skills/aris-run-experiment/SKILL.md
SKILL.md
Quality
Evals
Security

Quality

Content

67%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.

A highly actionable, well-sequenced runbook with real commands for every supported backend and explicit validation checkpoints. Its weaknesses are token efficiency — duplicated GPU checks and marketing-style setup prose — and inline content (W&B details, CLAUDE.md template, vendor setup) that would be better offloaded to reference files.

Suggestions

Remove the duplication between Step 0's inline compute guard and Step 2's pre-flight check — keep one canonical GPU-availability check and reference it from both places.

Trim the Vast.ai/Modal setup blockquotes (pricing, free-tier details, "ideal for users..." advice) to the minimum commands needed, or move them to a separate setup reference file.

Move the W&B integration details and the CLAUDE.md example into references/ files, keeping SKILL.md as a lean overview that links to them.

DimensionReasoningScore

Conciseness

The body is command-driven and largely avoids teaching known concepts, but the GPU-availability check is duplicated between Step 0 and Step 2, and the setup blockquotes contain advisory padding (free-tier pricing, "ideal for users without a local GPU"). This fits 'mostly efficient but includes some unnecessary explanation or could be tightened' rather than the efficient anchor 4.

3 / 5

Actionability

Provides copy-paste-ready ssh/screen/rsync/conda commands with consistent placeholders across all four environments (remote, Vast.ai, Modal, local), matching 'mostly executable guidance'. Falls short of anchor 5 because the W&B snippet uses pseudocode (config={...hyperparams...}) and local launch verification is vague ("Check process is running").

4 / 5

Workflow Clarity

Steps 0–7 are clearly sequenced with explicit checkpoints (Step 0 compute guard with a hard STOP, Step 5 verify launch, destroy only after results are collected) plus a Key Rules checklist. Validation is present, so the destructive/batch cap does not apply, but there is no feedback loop for a failed launch verification, keeping it below anchor 5.

4 / 5

Progressive Disclosure

No bundle files exist; all content sits in one well-sectioned file with clear headers and clearly signaled one-level references to sub-skills (/aris-serverless-modal, /aris-vast-gpu). This matches 'good structure; most content appropriately placed; minor organization gaps', though the ~320-line body inlines the CLAUDE.md template, W&B integration details, and vendor setup guides that could live in separate reference files.

4 / 5

Total

15

/

20

Passed

Description

83%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.

A strong description: it states a concrete, multi-environment capability and gives explicit bilingual trigger phrases for when to use it. The only weakness is that the action list is limited to deploy/run, omitting the skill's supporting capabilities.

DimensionReasoningScore

Specificity

Concrete actions ("Deploy and run ML experiments") are named along with four specific environments (local, remote, Vast.ai, Modal serverless GPU), matching the 'several specific actions; minor gaps in coverage' anchor. It falls short of a 5 because the supporting actions the skill actually performs (code sync, W&B logging, notifications, instance auto-destroy) are not mentioned.

4 / 5

Completeness

Explicitly answers both what ("Deploy and run ML experiments on local, remote, Vast.ai, or Modal serverless GPU") and when ("Use when user says 'run experiment', 'deploy to server', '跑实验', or needs to launch training jobs") with concrete trigger phrases, mirroring the anchor-5 example structure.

5 / 5

Trigger Term Quality

Includes natural phrases users would actually say — "run experiment", "deploy to server", "launch training jobs", and the Chinese "跑实验" — giving good keyword coverage. Not a 5 because common synonyms like "train a model" or "rent a GPU" are missing.

4 / 5

Distinctiveness Conflict Risk

A clear ML/GPU experiment deployment niche with distinct triggers, so mostly distinct with minimal conflict risk. "deploy to server" is slightly generic and could overlap with general deployment skills, keeping it below anchor 5.

4 / 5

Total

17

/

20

Passed

Validation

87%

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

CriteriaDescriptionResult

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
OpenLAIR/dr-claw
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.