Skill flagged — suspicious patterns detected
ClawHub Security flagged this skill as suspicious. Review the scan results before using.
ComfyUI
v1.0.1Run local ComfyUI workflows via the HTTP API. Use when the user asks to run ComfyUI, execute a workflow by file path/name, or supply raw API-format JSON; supports the default workflow bundled in assets.
⭐ 1· 3.1k·31 current·37 all-time
MIT-0
Download zip
LicenseMIT-0 · Free to use, modify, and redistribute. No attribution required.
Security Scan
OpenClaw
Benign
high confidencePurpose & Capability
Name/description, the run script, the download script, and the workflow assets all align with 'run ComfyUI workflows locally'. Required binary (python3) and references to a local ComfyUI install are appropriate for the stated purpose.
Instruction Scope
SKILL.md instructs the agent to read and edit workflow JSON (assets or user-supplied) and to run the bundled runner and downloader. This requires reading/writing files under the skill assets and interacting with the user's ~/ComfyUI install and its API on localhost — which is expected — but users should be aware the agent will make filesystem changes (write tmp workflow JSON) and may start/expect the ComfyUI server.
Install Mechanism
The skill is instruction-only (no global install spec), but the download_weights script can auto-download the pget binary from GitHub releases (https://github.com/replicate/pget/releases/latest/download) into ~/.local/bin and will write model files into ~/ComfyUI/models/<subfolder> as requested by the user. Using GitHub releases is a reasonable source, but it does download and write an executable to the user's home directory which the user may wish to review.
Credentials
The skill declares no environment variables or credentials. Scripts only use local paths (~/ComfyUI, ~/.local/bin) and network access when explicitly asked to download model weights or to call remote URLs provided by the user. No unrelated secrets are requested.
Persistence & Privilege
always:false and no modifications to other skills are requested. The only persistent changes possible are creating ~/.local/bin/pget (when downloading pget) and writing model files under the user's ComfyUI models directory — both are scoped to the user's home and are consistent with the skill's purpose.
Assessment
This skill appears coherent for running ComfyUI locally, but review these points before installing or running it:
- File writes: the downloader will write model files into ~/ComfyUI/models/<subfolder> and may create ~/.local/bin/pget (an executable downloaded from GitHub releases). If you run downloads, only provide trusted model URLs and consider running the downloader interactively so you can review what it will fetch.
- Executable install: pget will be downloaded to your home directory and made executable. If you prefer, install pget yourself or run the downloader with --no-pget to avoid automatic installation.
- Local server access: the run script communicates with a ComfyUI server expected at 127.0.0.1:8188. Make sure that server is a trusted local ComfyUI instance. The skill does not contact external APIs except for downloading files you explicitly provide.
- Workflow edits: the agent is instructed to modify workflow JSON (prompts, seeds, etc.) before running. Inspect any edited workflow (tmp file in the skill assets) if you have content-policy concerns — e.g., the bundled tmp-workflow contains an explicit sexual prompt.
If you are comfortable with the above (downloading model weights only from sources you trust and allowing the skill to write into your ComfyUI install), this skill is proportionate to its stated purpose.Like a lobster shell, security has layers — review code before you run it.
latestvk974pxc65dsn3rqjc6kn4xt7tx80jeb7
License
MIT-0
Free to use, modify, and redistribute. No attribution required.
Runtime requirements
🖼️ Clawdis
Binspython3