Pans Gpu Solution Builder

GPU 算力方案构建器。根据客户的工作负载类型（训练/推理/微调/批量推理/开发测试）、 部署规模（小型/中型/大型/企业级）和月度预算，自动推荐最优 GPU 配置 （H100/A100/L40S/A10G/RTX 4090 等），生成含 GPU 型号、数量、算力、 月度/小时成本的配置表、扩展路线图和技术建议，...

Security Scan

The skill's code and runtime instructions are self-contained and consistent with its description: it is a local Python-based GPU configuration recommender that does not request credentials, external installs, or network access.

Purpose & Capability

Name and description (GPU solution builder) match the included script and SKILL.md examples. The script computes GPU recommendations, costs, and scaling roadmaps — all coherent with the stated purpose. Nothing requested (no env vars, no binaries) appears disproportionate.

Instruction Scope

SKILL.md instructs running the included Python script with command-line arguments; the script only reads those arguments and internal data structures. There are no instructions to read unrelated files, access credentials, or send data externally.

Install Mechanism

No install spec or external downloads. Code is provided as a local Python script that uses only the standard library, so no external package installation is required.

Credentials

The skill declares no required environment variables, credentials, or config paths and the code does not access environment secrets. Requested privileges are proportional to the stated functionality.

Persistence & Privilege

always is false and the skill is user-invocable. It doesn't attempt to modify other skills or system-wide settings. Autonomous invocation is enabled by default but is not combined with other red flags here.

Assessment

This is a local, self-contained Python script that generates GPU configuration recommendations. Before installing or running: (1) review the full script yourself (it appears safe and uses only the Python standard library); (2) run it in an isolated/sandbox environment if you don't fully trust the source; (3) verify cost numbers and assumptions — the tool uses fixed example prices and heuristics and should not be treated as authoritative for procurement; (4) ensure you have Python 3 available and that you run the provided commands rather than arbitrary variants. If you want additional assurance, ask the publisher for provenance or run a static check of the full file contents.

Overview

GPU 算力方案构建器。根据客户需求自动构建 GPU 算力方案。

工作负载类型

类型	说明	推荐 GPU
training	大语言模型训练、深度学习训练	H100, A100 80GB
inference	在线推理服务、API 部署	L40S, A10G, A100
finetuning	LoRA/QLoRA 微调、领域适配	A100 40GB, L40S
inference_batch	离线批量处理、数据标注	RTX 4090, A10G
development	模型开发、实验调试	RTX 4090, A10G

规模定义

规模	说明	最大预算
small	起步/实验阶段	$5,000/月
medium	生产环境/中等负载	$20,000/月
large	大规模训练/推理	$100,000/月
enterprise	超大规模/集群部署	$500,000/月

使用方法

# 生成训练方案（中规模，$10,000预算）
python3 scripts/solution.py --workload training --budget 10000 --scale medium

# 生成推理方案（小型，$5,000预算）
python3 scripts/solution.py --workload inference --budget 5000 --scale small --format json

# 生成微调方案，输出到文件
python3 scripts/solution.py --workload finetuning --budget 8000 --scale medium --output solution.md

输出内容

• GPU 配置推荐表：型号、数量、总显存、算力、月成本、预算匹配评分
• 成本估算：月度、小时、年度预估
• 扩展路线图：从当前规模到更大规模的升级路径
• 技术建议：网络配置、存储方案、分布式训练框架、推理优化建议