← Back to Skills Marketplace
axelhu

Axelhu Local Sdxl

by AxelHu · GitHub ↗ · v1.0.1 · MIT-0
cross-platform ⚠ suspicious
94
Downloads
0
Stars
0
Active Installs
2
Versions
Install in OpenClaw
/install axelhu-local-sdxl
Description
基于本地 ComfyUI 与 SDXL,实现高质量私密图像生成,支持详细构图、风格及尺寸自定义,适合无在线API场景。
README (SKILL.md)

本地 SDXL 生图 (axelhu-local-sdxl)

基于 ComfyUI + SDXL 的本地图片生成技能。适用于需要高质量配图、素材生成、本地私密生图、或对在线生成有频率限制的场景。

触发时机

在以下场景时使用:

  • 用户/其他 agent 说"本地生图"、"本地SD生图"、"用本地工具生成图片"
  • 用户/其他 agent 说"私密生图"、"脱敏生图"、"不上传在线"
  • 需要生成涉及隐私、测试、商业素材的内容,且明确指定本地处理
  • 需要精细控制构图、风格、尺寸的生图任务,且指定本地执行

不要用于:

  • 默认生图请求(那些默认走在线服务,如 Midjourney/DALL-E)
  • 用户说"生成图片"、"画一张图"且没有指定本地(默认在线优先)

使用方式

直接描述想要的图片内容,例如:

  • "帮我生成一张科技感十足的封面图"
  • "画一个程序员深夜编程的场景,插画风格"
  • "生成一张杭州西湖的夜景图,高清写实风格"

Agent 收到后执行生图脚本,发送结果给用户。

技术规格

项目 参数
模型 Stable Diffusion XL 1.0 (fp16)
显卡 NVIDIA RTX 3080 (10GB VRAM)
出图尺寸 默认 1024×768,可调整
生图速度 ~18秒/张(euler 采样,20步)
可用采样器 euler, dpmpp_2m, lcm, ddim 等
API 端口 localhost:8188

Agent 调用方法

方式一:通过脚本调用(推荐)

import subprocess
result = subprocess.run(
    ["python3", "/path/to/scripts/sdxl_generate.py",
     "--prompt", "一只狐狸在森林里",
     "--negative", "模糊, 低质量",
     "--steps", "20",
     "--seed", "42",
     "--output", "/tmp/output.png"],
    capture_output=True, text=True
)
print(result.stdout)

方式二:直接调 ComfyUI REST API

import requests, time, json

COMFYUI = "http://localhost:8188"

def generate(prompt, negative="", steps=20, seed=42, width=1024, height=768):
    workflow = {
        "loader": {"class_type": "CheckpointLoaderSimple",
                   "inputs": {"ckpt_name": "sdxl-base-1.0.safetensors"}},
        "positive": {"class_type": "CLIPTextEncode",
                     "inputs": {"text": prompt, "clip": ["loader", 1]}},
        "negative": {"class_type": "CLIPTextEncode",
                      "inputs": {"text": negative, "clip": ["loader", 1]}},
        "latent": {"class_type": "EmptyLatentImage",
                   "inputs": {"width": width, "height": height, "batch_size": 1}},
        "sampler": {"class_type": "KSampler",
                    "inputs": {"seed": seed, "steps": steps, "cfg": 7.0,
                               "sampler_name": "euler", "scheduler": "normal",
                               "denoise": 1.0,
                               "positive": ["positive", 0], "negative": ["negative", 0],
                               "model": ["loader", 0], "latent_image": ["latent", 0]}},
        "decode": {"class_type": "VAEDecode",
                   "inputs": {"samples": ["sampler", 0], "vae": ["loader", 2]}},
        "save": {"class_type": "SaveImage",
                 "inputs": {"images": ["decode", 0], "filename_prefix": "sdxl_gen"}}
    }
    r = requests.post(f"{COMFYUI}/prompt", json={"prompt": workflow})
    pid = r.json()["prompt_id"]
    # 等待完成(轮询)
    for _ in range(30):
        time.sleep(2)
        hist = requests.get(f"{COMFYUI}/history/{pid}").json()
        if pid in hist:
            return f"{COMFYUI}/view?filename=sdxl_gen_00001_.png"
    return None

参数说明

参数 默认值 说明
--prompt (必填) 图片描述,越详细越好
--negative "blurry, low quality, distorted" 反向提示词
--steps 20 采样步数,越高质量越好越慢
--seed 随机 种子,决定具体画面
--width 1024 宽度(8的倍数)
--height 768 高度(8的倍数)
--sampler euler 采样器(euler/dpmpp_2m/lcm/ddim)
--cfg 7.0 CFG 强度(1-20)
--output 自动 输出路径

快速参考:常用场景

场景 推荐参数
写实风景/人物 --steps 25 --cfg 7 --sampler euler
动漫/插画风格 --steps 20 --cfg 8 --negative "写实, 照片"
快速草图/验证 --steps 4 --sampler lcm --cfg 1.0(需要 LCM 模型)
16:9 横幅图 --width 1280 --height 720

输出格式

生图完成后,脚本自动将图片保存到指定路径或 /tmp/ 目录,并通过飞书发送图片消息给用户。

注意事项

  • ComfyUI 服务必须已启动(python3 main.py --port 8188
  • RTX 3080 推荐分辨率 1024×768 以下,更高分辨率可能爆显存
  • 生图过程约 18 秒,请提前告知用户等待
  • 图片风格默认写实,如需特定风格在 prompt 中描述
Usage Guidance
This skill appears to be a local ComfyUI SDXL helper and mostly does what it says, but there are a few things to check before installing or running it: 1) The SKILL.md mentions automatically sending results via Feishu, but neither script contains any Feishu or external messaging code — expect to handle delivery yourself or add secure messaging code (and then securely store credentials). 2) The Python script uses hardcoded paths (/home/axelhu/ComfyUI/output/) and will fail or read unexpected files on other systems — update the script to use configurable output directories before running. 3) Ensure ComfyUI is only listening on localhost (127.0.0.1) as recommended to avoid remote access, and confirm model files are present locally (large download). 4) Review and run the scripts as a non-root user, and inspect the code (simple HTTP calls to localhost and local filesystem operations) to satisfy yourself there is no data exfiltration. If you need the Feishu/send behavior, require that the skill request explicit credentials and implement them transparently rather than claiming the capability without code.
Capability Analysis
Type: OpenClaw Skill Name: axelhu-local-sdxl Version: 1.0.1 The skill bundle provides a legitimate toolset for local image generation using ComfyUI and SDXL. The Python script `scripts/sdxl_generate.py` and shell script `scripts/start-comfyui.sh` are transparently written to manage local processes and interact with a local REST API (localhost:8188). The instructions in `SKILL.md` are well-defined, providing clear triggers and usage examples without any signs of prompt injection, data exfiltration, or unauthorized execution. While it contains hardcoded paths specific to the author's environment (e.g., `/home/axelhu/`), there is no evidence of malicious intent.
Capability Assessment
Purpose & Capability
The name/description (local ComfyUI + SDXL) aligns with the included scripts and REST examples. However the SKILL.md claims the script will "通过飞书发送图片消息给用户" (send results via Feishu), but no code implements Feishu or any external messaging — a discrepancy between claimed capabilities and actual implementation. The script also hardcodes paths and filenames under /home/axelhu which is specific to the author's environment and may not work for other users.
Instruction Scope
Runtime instructions are focused on running the local script or calling ComfyUI's REST API on localhost:8188, which is appropriate. The script only talks to localhost and the ComfyUI API and reads local output directories. It does, however, read a hardcoded output directory (/home/axelhu/ComfyUI/output/) rather than using a configurable path in several places, which is scope-coupled to the author's environment and could cause unexpected file reads on install.
Install Mechanism
There is no install spec (instruction-only skill) and the files are small scripts and docs. No download-from-URL or package installation is defined by the skill itself. The deployment notes correctly describe a large local model requirement (sdxl-base-1.0.safetensors ~6.5GB), which is expected for the stated purpose.
Credentials
The skill requests no environment variables or credentials, which is proportionate — except SKILL.md claims automatic sending via Feishu (which would normally require credentials). That capability is asserted but not implemented in the code and yet no env vars/credentials are requested. The hardcoded filesystem paths (/home/axelhu/...) are a privacy/permission concern because the script enumerates and copies files from that directory.
Persistence & Privilege
Skill does not ask for persistent privileges; always=false and no modifications to other skills or system-wide settings. The script runs only on-demand and only communicates with localhost:8188 (ComfyUI) — no autonomous remote callbacks are present.
How to Use
  1. Make sure OpenClaw is installed (local or Docker)
  2. Run the install command in chat: /install axelhu-local-sdxl
  3. After installation, invoke the skill by name or use /axelhu-local-sdxl
  4. Provide required inputs per the skill's parameter spec and get structured output
Version History
v1.0.1
Version 1.0.1 - 调整了触发条件,明确要求需用户/agent 明确指定"本地"、"私密"等关键词才会调本地生图 - 默认情况及普通"生成图片"请求将优先走在线服务,不再自动触发本地生图 - 新增部署说明文档 references/deployment.md - 新增 ComfyUI 启动脚本 scripts/start-comfyui.sh
v1.0.0
- Initial release of axelhu-local-sdxl: a local image generation skill based on ComfyUI + SDXL. - Supports high-quality image, material, and private generation scenarios without online API limits. - Triggered when users request image creation without specifying online models. - Provides detailed usage instructions, script/API integration methods, and full parameter guide. - Output images locally and send results via Feishu.
Metadata
Slug axelhu-local-sdxl
Version 1.0.1
License MIT-0
All-time Installs 0
Active Installs 0
Total Versions 2
Frequently Asked Questions

What is Axelhu Local Sdxl?

基于本地 ComfyUI 与 SDXL,实现高质量私密图像生成,支持详细构图、风格及尺寸自定义,适合无在线API场景。 It is an AI Agent Skill for Claude Code / OpenClaw, with 94 downloads so far.

How do I install Axelhu Local Sdxl?

Run "/install axelhu-local-sdxl" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.

Is Axelhu Local Sdxl free?

Yes, Axelhu Local Sdxl is completely free, licensed under MIT-0. You can download, install and use it at no cost.

Which platforms does Axelhu Local Sdxl support?

Axelhu Local Sdxl is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).

Who created Axelhu Local Sdxl?

It is built and maintained by AxelHu (@axelhu); the current version is v1.0.1.

💬 Comments