← 返回 Skills 市场
98
总下载
0
收藏
0
当前安装
1
版本数
在 OpenClaw 中安装
/install gemmamatch
功能描述
Auto-detect hardware and recommend the best Gemma 4 model for local deployment on PC, Mac, or mobile.
使用说明 (SKILL.md)
GemmaMatch — Gemma 4 Local Hardware Matcher
Find the best Gemma 4 model for your hardware in seconds.
Website: https://www.gemmamatch.com
What it does
GemmaMatch auto-detects your GPU, VRAM, and system specs via WebGPU/WebGL APIs, then recommends the most suitable Gemma 4 model tier and provides a ready-to-use run command. All processing happens locally in your browser — no data leaves your device.
Recommended model tiers
| Tier | Target hardware | Use case |
|---|---|---|
| Gemma 4 E2B | Phones, tablets, low-VRAM devices | On-device inference, edge deployment |
| Gemma 4 26B MoE | Desktop GPUs (8-16 GB VRAM) | General local AI, coding assistance |
| Gemma 4 31B Dense | Workstations (24+ GB VRAM) | High-quality generation, research |
Key features
- Automatic GPU detection — uses WebGPU and WebGL APIs, no install required
- Personalized model recommendation — matches your exact hardware to the optimal Gemma 4 variant
- Platform-specific setup guides — step-by-step instructions for Mac (MLX, Ollama), Windows (Ollama, LM Studio), iOS, and Android
- One-click run commands — get a copy-paste Ollama or LM Studio command tailored to your system
- Manual comparison mode — compare upgrade scenarios or override auto-detection
- Privacy-first — everything runs in-browser, zero data collection
Quick start
- Visit https://www.gemmamatch.com
- Allow hardware detection (or enter specs manually)
- Get your recommended model + run command
- Copy the command and run it in your terminal
Supported platforms
- macOS — Apple Silicon (M1-M4), Intel with discrete GPU
- Windows — NVIDIA (RTX 30/40/50 series), AMD (RX 7000 series)
- Linux — NVIDIA CUDA, AMD ROCm
- iOS / Android — on-device model recommendations
Links
安全使用建议
This skill appears coherent, but exercise normal caution before acting on any recommended terminal commands: 1) Confirm the website uses HTTPS and inspect the GitHub source (https://github.com/walex8925/Gemma4local) if you can. 2) Review any copy-paste run commands before executing — avoid piping unknown scripts into a shell (e.g., curl | sh). 3) Verify the site’s privacy claim by checking whether detection scripts make network calls (you can inspect the page or the repo). 4) If you want complete assurance, open the repository code locally or in a sandboxed browser to validate that hardware detection runs purely in-browser and no external data is exfiltrated.
功能分析
Type: OpenClaw Skill
Name: gemmamatch
Version: 1.0.0
The skill bundle contains only metadata and documentation (SKILL.md) describing a hardware matching service for LLMs. There is no executable code, no scripts, and no instructions directing the AI agent to perform any system-level actions or data exfiltration. The content is purely informational and promotional, pointing to external websites and repositories.
能力评估
Purpose & Capability
The name and description (auto-detect hardware and recommend Gemma 4 variants) align with the SKILL.md content. The skill is instruction-only and uses browser WebGPU/WebGL detection and platform-specific run commands — these are coherent with the stated purpose.
Instruction Scope
SKILL.md instructs users to visit the website and allow in-browser hardware detection; it does not instruct the agent to read system files, env vars, or send data elsewhere. Note: detection happens in the user's browser (client-side) rather than the agent, so the skill relies on the website and user consent to run detection.
Install Mechanism
No install spec or code files are present (instruction-only), so nothing is written to disk or downloaded by the skill itself.
Credentials
The skill requests no environment variables, credentials, or config paths — this is proportionate to a read-only hardware-detection/recommendation guide.
Persistence & Privilege
The skill is not always-on and uses default model invocation settings; it does not request elevated or persistent system privileges.
如何使用
- 确保已安装 OpenClaw（本地或 Docker 部署）
- 在对话框中输入安装命令：
/install gemmamatch - 安装完成后，直接呼叫该 Skill 的名称或使用
/gemmamatch触发 - 根据 Skill 的参数说明提供必要输入，即可获得结构化输出
版本历史
v1.0.0
Initial release: auto-detect hardware and recommend the best Gemma 4 model for local deployment.
元数据
常见问题
GemmaMatch — Gemma 4 Local Hardware Matcher 是什么？
Auto-detect hardware and recommend the best Gemma 4 model for local deployment on PC, Mac, or mobile. 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件，目前累计下载 98 次。
如何安装 GemmaMatch — Gemma 4 Local Hardware Matcher？
在 OpenClaw 或 Claude Code 对话框中运行命令「/install gemmamatch」即可一键安装，无需额外配置。
GemmaMatch — Gemma 4 Local Hardware Matcher 是免费的吗？
是的，GemmaMatch — Gemma 4 Local Hardware Matcher 完全免费，采用 MIT-0 许可证，可自由下载、安装和使用。
GemmaMatch — Gemma 4 Local Hardware Matcher 支持哪些平台？
GemmaMatch — Gemma 4 Local Hardware Matcher 跨平台运行，可在任意部署了 OpenClaw / Claude Code 的环境中使用（cross-platform）。
谁开发了 GemmaMatch — Gemma 4 Local Hardware Matcher？
由 walex8925（@walex8925）开发并维护，当前版本 v1.0.0。
推荐 Skills