← 返回 Skills 市场
MFlux Skill for OpenClaw
作者
Pankaj Jain
· GitHub ↗
· v0.1.0
384
总下载
0
收藏
1
当前安装
1
版本数
在 OpenClaw 中安装
/install mflux
功能描述
Local image generation using Apple MLX via mflux — FLUX.2 Klein 4B (fast, Apache 2.0) and Z-Image Turbo (quality) models
安全使用建议
This skill appears coherent, but before installing: (1) verify the origin and integrity of the mflux package (PyPI project page or official repo) rather than blindly running installs; (2) confirm what 'uv' is and prefer a controlled environment (virtualenv) when using pip; (3) be prepared for large model downloads and ~8–32GB+ storage usage (models cached in ~/.cache/huggingface/hub/); (4) some models or LoRA files may be third‑party or require a Hugging Face token for private repos—only provide tokens if you trust the model source; (5) review licenses of models you plan to use (SKILL.md lists some as 'Custom'); and (6) avoid using LoRA/safetensors from untrusted sources. If you want higher assurance, ask the publisher for a source repository or official homepage and inspect the mflux project directly before installation.
功能分析
Type: OpenClaw Skill
Name: mflux
Version: 0.1.0
The mflux skill bundle provides standard documentation and instructions for local image generation on Apple Silicon using the legitimate mflux library. The SKILL.md and README.md files contain helpful usage examples, installation steps via uv/pip, and parameter references without any evidence of malicious intent, data exfiltration, or prompt injection attacks.
能力评估
Purpose & Capability
Name/description (local image generation on Apple Silicon) match the instructions: install mflux (pip/uv) and run mflux CLI or Python API. Required platform and RAM guidance are appropriate for the claimed capabilities.
Instruction Scope
SKILL.md stays on‑topic (installing and using mflux, model selection, quantization, LoRA, image-to-image workflows). It does implicitly require network access to download models and caches them in ~/.cache/huggingface/hub/ and may require a Hugging Face token for private models — this is expected but worth noting.
Install Mechanism
No install spec in the registry; SKILL.md recommends installing via uv (a third‑party tool) or pip. No direct downloads from unknown URLs or archive extraction are included in the skill files themselves.
Credentials
The skill declares no required environment variables or credentials. The instructions do not request secrets. A possible real‑world need for HF credentials (only for private Hugging Face models) is noted in the SKILL.md behavior and is proportionate to model downloads.
Persistence & Privilege
Skill does not request always:true, does not claim persistent elevated privileges, and is user‑invocable. It does instruct installing third‑party software (mflux) which will persist as any pip/uv install would.
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 mflux - 安装完成后，直接呼叫该 Skill 的名称或使用
/mflux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
v0.1.0
Initial release of mflux: local image generation on Apple Silicon with Apache 2.0 and high-quality models.
- Generate images locally using Apple MLX on Mac (M1 or later), no cloud/API keys needed
- Supports fast FLUX.2-klein-4B (default, Apache 2.0), FLUX.2-klein-9B, Z-Image-Turbo, and more
- Optional low-RAM mode via quantization (8-bit, 4-bit)
- Python API and CLI for image, img2img, LoRA, and editing workflows
- Automatic model downloading/caching, with RAM/speed/quality comparison tables for optimal use
元数据
常见问题
MFlux Skill for OpenClaw 是什么？
Local image generation using Apple MLX via mflux — FLUX.2 Klein 4B (fast, Apache 2.0) and Z-Image Turbo (quality) models. 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件，目前累计下载 384 次。
如何安装 MFlux Skill for OpenClaw？
在 OpenClaw 或 Claude Code 对话框中运行命令「/install mflux」即可一键安装，无需额外配置。
MFlux Skill for OpenClaw 是免费的吗？
是的，MFlux Skill for OpenClaw 完全免费（开源免费），可自由下载、安装和使用。
MFlux Skill for OpenClaw 支持哪些平台？
MFlux Skill for OpenClaw 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 MFlux Skill for OpenClaw？
由 Pankaj Jain（@pjain）开发并维护，当前版本 v0.1.0。
推荐 Skills