Name: flox-cuda
Rating: 60 (1 reviews)
Author: flox

flox-cuda

CUDA and GPU development with Flox. Use for NVIDIA CUDA setup, GPU computing, deep learning frameworks, cuDNN, and cross-platform GPU/CPU development.

Quality

70%

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

Fix and improve this skill with Tessl

tessl review fix ./flox-plugin/skills/flox-cuda/SKILL.md

Quality

Content

57%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.

Highly actionable with copy-paste-ready CUDA setup examples, but the body is verbose due to heavy repetition of identical command and TOML blocks, and all content is inlined with no progressive disclosure into reference files. Workflow structure is topical rather than a validated multi-step sequence.

Suggestions

De-duplicate the repeated flox search commands and the cuda_nvcc/cuda_cudart/systems TOML block by defining a canonical base config once and referencing it, to cut significant token weight.

Split the package-pattern table and the full per-framework environment examples into a reference file under references/ and link to it from SKILL.md, turning the overview into a lean entry point.

Add an explicit ordered setup workflow with validation checkpoints (install → verify with nvcc --version/nvidia-smi → compile hello_cuda.cu → only then proceed), with a fix→retry loop tied to the Troubleshooting section.

Dimension	Reasoning	Score
Conciseness	The ~530-line body is noticeably padded: 'Core Commands' and 'Package Discovery' duplicate the same search commands verbatim, and the full cuda_nvcc/cuda_cudart/systems TOML block is repeated across Basic, PyTorch, TensorFlow, Multi-GPU, and Modular sections. Not a 1 because it avoids explaining basic concepts, but the repeated near-identical config blocks are clear padding.	2 / 5
Actionability	Provides fully executable, copy-paste-ready TOML and bash including a complete hello_cuda.cu test program and PyTorch/TensorFlow setups, covering the common cases concretely.	5 / 5
Workflow Clarity	Content is organized topically (Prerequisites → Commands → Examples → Testing → Troubleshooting) with verification steps present, but these are scattered reference sections rather than an explicit sequenced workflow with validation checkpoints and a fix→retry loop.	3 / 5
Progressive Disclosure	No bundle files exist and there are no external file references; all reference material (package tables, many full environment configs, debugging tools) is inlined in a single document. Section headers provide some structure, but bulk content that could be split into reference files is inlined with no one-level-deep navigation.	3 / 5
	Total	13 / 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, third-person description that explicitly pairs a clear 'what' with a concrete 'Use for...' trigger list covering the main CUDA/GPU scenarios. It is specific and distinctive, with only minor gaps in keyword synonym coverage.

Dimension	Reasoning	Score
Specificity	Names the CUDA/GPU domain plus several concrete capability areas ('NVIDIA CUDA setup', 'GPU computing', 'deep learning frameworks', 'cuDNN', 'cross-platform GPU/CPU development'); falls short of 5 because it lists domains rather than discrete executable actions like extract/fill/merge.	4 / 5
Completeness	Explicitly states both what ('CUDA and GPU development with Flox') and when ('Use for NVIDIA CUDA setup, GPU computing, deep learning frameworks, cuDNN...') with concrete trigger phrases, matching the anchor for clearly answering both what and when.	5 / 5
Trigger Term Quality	Includes natural developer-facing terms ('NVIDIA CUDA setup', 'GPU computing', 'cuDNN', 'deep learning frameworks') but misses common synonyms and product names a user might say (e.g., PyTorch, TensorFlow, nvcc), keeping it just below comprehensive.	4 / 5
Distinctiveness Conflict Risk	The CUDA/GPU specialization carves a clear niche distinct from generic skills, with only minor overlap risk against the sibling Flox skills it itself references.	4 / 5
	Total	17 / 20 Passed

Validation

93%

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 — 15 / 16 Passed

Validation for skill structure

Criteria	Description	Result
skill_md_line_count	SKILL.md is long (533 lines); consider splitting into references/ and linking	Warning

	Total	15 / 16 Passed

Repository: flox/flox-agentic
Path: flox-plugin/skills/flox-cuda/SKILL.md
Commit: aaa5f1a

Reviewed: 15 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.