A specialized automation toolkit for setting up and optimizing ComfyUI on AMD GPUs using Windows DirectML.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install comfyui-directml-amd
2. Manual Installation
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
3. Prompt Installation
Copy this prompt to OpenClaw to install it automatically.
Help me install comfyui-directml-amd using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Prefer to download?
Get the raw skill files in a ZIP archive.
What is ComfyUI DirectML AMD Skill?
The ComfyUI DirectML AMD Skill provides a robust environment for running high-performance AI generation workflows on AMD hardware. While many AI tools are optimized for CUDA, this skill utilizes Microsoft DirectML to ensure that Radeon RX 7000, 6000, and 5000 series cards can execute complex image, music, and video generation models locally.
By leveraging Openclaw Skills, users can automate the tedious process of patching source files like model_patcher.py and managing specific Python 3.12 virtual environments. This skill transforms a manual, error-prone setup into a streamlined developer experience, ensuring compatibility with SDXL, Flux.1, and other state-of-the-art models.
ComfyUI DirectML AMD Skill Use Cases
- Running local AI image generation on Windows with AMD Radeon GPUs.
- Resolving recurring torch_directml module errors in automated workflows.
- Patching ComfyUI core files to prevent model loading syntax errors.
- Benchmarking VRAM usage and generation speed across different SDXL models.
- Deploying headless ComfyUI instances for use within Openclaw Skills integrations.
How ComfyUI DirectML AMD Skill Works
- The skill initializes a dedicated Python 3.12 environment, bypassing compatibility issues with newer Python versions.
- It triggers a force-reinstall of the DirectML torch backend to ensure hardware acceleration is active.
- An automated patching script modifies the model_patcher.py file within ComfyUI to handle DirectML-specific tensor operations.
- Configuration flags are dynamically applied based on the user's available VRAM (8GB, 12GB, or 16GB+).
- The system establishes a connection to the Openclaw Skills ecosystem via the comfy plugin for workflow automation.
ComfyUI DirectML AMD Skill Setup
# Install via the hub
clawhub install comfyui-directml-amd
# Prepare the environment
cd C:\ComfyUI
uv venv --python 3.12
# Install DirectML dependencies
.venv\Scripts\python.exe -m pip install torch-directml --force-reinstall
# Apply the automated fixes
python apply-directml-fixes.py
# Launch the optimized server
.venv\Scripts\python.exe main.py --directml --port 8188
ComfyUI DirectML AMD Skill Data Schema & Taxonomy
| Component | Function | Data Handled |
|---|---|---|
| apply-directml-fixes.py | Patching Script | Modifies model_patcher.py for compatibility |
| download-models.ps1 | Asset Manager | Orchestrates downloads of SDXL and Flux models |
| benchmark.py | Analytics Tool | Logs generation times and VRAM usage metrics |
| OpenClaw Config | Integration | Maps local port 8188 to the agent workflow plugin |
ComfyUI DirectML AMD Skill Advanced Features
- Automated source code modification with built-in backup and rollback functionality.
- Support for FP8 and low-VRAM optimizations tailored specifically for AMD hardware.
- Integrated performance benchmarking comparing DirectML generation speeds against CUDA references.
- Seamless Openclaw Skills integration for programmatically triggering workflows from an AI agent.
SKILL.md
Loading
Related Openclaw Skills
Thumbnail Designer
An AI-driven design engine specialized in crafting high-conversion YouTube and social media thumbnails using psychology and A/B testing frameworks.
Tooling Engineer
A senior-level AI skill focused on architecting, implementing, and optimizing high-performance developer tools, CLI applications, and extensible plugin systems.
Test Automator
A senior test automation expert focused on building scalable frameworks, CI/CD integration, and high-coverage automated testing solutions.
Terraform Engineer
An expert AI agent skill for designing, implementing, and managing enterprise-grade infrastructure as code using Terraform across multi-cloud environments.
Agentic Lending EVK Read-Only
A non-executing planning skill for resolving oracle routes and preparing Euler Vault Kit (EVK) lending market artifacts.
Agentic Lending Morpho Read-only
A specialized skill for read-only planning and readiness analysis of Api3-backed Morpho lending markets.
Featured*
