← Back to Skills Marketplace
98
Downloads
0
Stars
0
Active Installs
1
Versions
Install in OpenClaw
/install gemmamatch
Description
Auto-detect hardware and recommend the best Gemma 4 model for local deployment on PC, Mac, or mobile.
README (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
Usage Guidance
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.
Capability Analysis
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.
Capability Assessment
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.
How to Use
- Make sure OpenClaw is installed (local or Docker)
- Run the install command in chat:
/install gemmamatch - After installation, invoke the skill by name or use
/gemmamatch - Provide required inputs per the skill's parameter spec and get structured output
Version History
v1.0.0
Initial release: auto-detect hardware and recommend the best Gemma 4 model for local deployment.
Metadata
Frequently Asked Questions
What is GemmaMatch — Gemma 4 Local Hardware Matcher?
Auto-detect hardware and recommend the best Gemma 4 model for local deployment on PC, Mac, or mobile. It is an AI Agent Skill for Claude Code / OpenClaw, with 98 downloads so far.
How do I install GemmaMatch — Gemma 4 Local Hardware Matcher?
Run "/install gemmamatch" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.
Is GemmaMatch — Gemma 4 Local Hardware Matcher free?
Yes, GemmaMatch — Gemma 4 Local Hardware Matcher is completely free, licensed under MIT-0. You can download, install and use it at no cost.
Which platforms does GemmaMatch — Gemma 4 Local Hardware Matcher support?
GemmaMatch — Gemma 4 Local Hardware Matcher is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).
Who created GemmaMatch — Gemma 4 Local Hardware Matcher?
It is built and maintained by walex8925 (@walex8925); the current version is v1.0.0.
More Skills