← Back to Skills Marketplace
Comfyskill
by
edward-hsiao
· GitHub ↗
· v1.0.3
· MIT-0
332
Downloads
0
Stars
1
Active Installs
4
Versions
Install in OpenClaw
/install comfyskill
Description
Send prompts to a local ComfyUI instance to generate images based on user descriptions.
README (SKILL.md)
ComfyUI 生成技能
描述
此技能可將提示詞發送至本地 ComfyUI 並生成圖片。
指令
generate_image(prompt: string): 接收使用者的描述並發送到 ComfyUI。
設定
- endpoint: http://127.0.0.1:8188
- workflow: workflow.json
Usage Guidance
This skill looks like a small ComfyUI helper but contains clear inconsistencies and coding errors. Before installing or using it: (1) verify and set WORKFLOW_PATH or update the code to use the included workflow.json (currently the code expects text2image-1.json in WORKFLOW_PATH); (2) fix the implementation to use the provided prompt argument (the code references an undefined prompt variable); (3) ensure COMFYUI_ENDPOINT is set to a trusted local address (default is localhost:8188) — otherwise the skill could send workflow data to a remote host; (4) review the workflow.json contents to ensure no sensitive local paths are exposed (it references a checkpoint filename). If you don't control the environment variables or cannot audit/patch the code, treat this skill as risky and avoid installing it.
Capability Analysis
Type: OpenClaw Skill
Name: comfyskill
Version: 1.0.3
The skill is designed to interface with a local ComfyUI instance for image generation. While the code contains functional bugs—specifically referencing an undefined 'prompt' variable (likely intended to be 'args.prompt') and looking for 'text2image-1.json' instead of the provided 'workflow.json'—there is no evidence of malicious intent. The use of 'fs.readFileSync' and 'fetch' to a local endpoint (127.0.0.1:8188) is consistent with the stated purpose of the skill in index.js and SKILL.md.
Capability Assessment
Purpose & Capability
The skill claims to send prompts to a local ComfyUI instance and includes a bundled workflow.json — that aligns with its purpose. However, the runtime code expects an environment variable WORKFLOW_PATH and a file named text2image-1.json (process.env.WORKFLOW_PATH + 'text2image-1.json'), while SKILL.md references workflow.json and lists no required env vars. This inconsistency means the code likely won't use the included workflow.json unless the environment is set up to match the code.
Instruction Scope
SKILL.md describes a simple generate_image(prompt) API and a default endpoint, but the actual index.js reads a local file (workflow) from an env path and POSTs workflow JSON to the endpoint. The SKILL.md does not declare that the agent must have WORKFLOW_PATH or that the code will read local files. Also the code uses an undefined variable prompt instead of the function argument, which is a logic bug and could cause runtime errors.
Install Mechanism
No install spec and only small code files are included. Nothing is downloaded or written during install — lowest install risk.
Credentials
The skill metadata declares no required environment variables, but index.js reads process.env.COMFYUI_ENDPOINT (with a sensible default) and process.env.WORKFLOW_PATH (no default) — the missing declaration for WORKFLOW_PATH is a mismatch. Also COMFYUI_ENDPOINT can be set to a non-local endpoint, which would cause the skill to send workflow/prompt data to a remote host if misconfigured.
Persistence & Privilege
The skill is not always-enabled and does not request elevated persistence or modify other skills. It simply reads files and posts to an endpoint at runtime.
How to Use
- Make sure OpenClaw is installed (local or Docker)
- Run the install command in chat:
/install comfyskill - After installation, invoke the skill by name or use
/comfyskill - Provide required inputs per the skill's parameter spec and get structured output
Version History
v1.0.3
- Bumped package version to 1.0.3 in package.json.
- No functional or documentation changes in SKILL.md.
v1.0.2
- Minor update with no user-facing changes.
- Internal code modified in index.js; SKILL.md unchanged.
v1.0.1
Initial release of comfyskill.
- Added core files: index.js and package.json.
- Provides the ability to generate images by sending prompts to a local ComfyUI instance.
- Exposes the generate_image(prompt: string) command.
v1.0.0
comfyskill v1.0.0
- 初始发布版本。
- 实现将用户提示词发送到本地 ComfyUI 并生成图片。
- 提供 generate_image 指令，支持描述性文本输入。
- 支持配置 endpoint 和 workflow 参数。
Metadata
Frequently Asked Questions
What is Comfyskill?
Send prompts to a local ComfyUI instance to generate images based on user descriptions. It is an AI Agent Skill for Claude Code / OpenClaw, with 332 downloads so far.
How do I install Comfyskill?
Run "/install comfyskill" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.
Is Comfyskill free?
Yes, Comfyskill is completely free, licensed under MIT-0. You can download, install and use it at no cost.
Which platforms does Comfyskill support?
Comfyskill is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).
Who created Comfyskill?
It is built and maintained by edward-hsiao (@edward-hsiao); the current version is v1.0.3.
More Skills