ZenMux Image Generation

Generate images via ZenMux API (Pro/Elite). Supports Text-to-Image, Image-to-Image, and Multi-Image reference fusion.

This skill appears to do what it claims, but take these precautions before installing or running it: - Note the metadata omission: the SKILL.md and scripts require ZENMUX_API_KEY, but the registry metadata does not list any required env vars. Don't provide credentials until you confirm where they will be used and stored. - Review the script yourself (it is short) and verify the outbound endpoint (https://zenmux.ai/api/vertex-ai). If you don't trust that domain or vendor, do not supply keys or run the script. - Run the script in an isolated environment (container or VM) and ensure Python dependencies (requests) are installed from trusted sources (pip from PyPI) before execution. - Only use API keys with least privilege and be mindful of sensitive input images (e.g., minors, IDs); multi-image fusion could combine private images in unexpected ways. - Contact the skill publisher or maintainers to ask them to update registry metadata to declare ZENMUX_API_KEY as a required credential and to provide installation/dependency instructions. If you need lower risk: ask for a version that includes a proper install spec (dependencies declared) and correct required-env metadata before enabling.

Type: OpenClaw Skill Name: zenmux-image-generation Version: 1.5.0 The skill bundle is benign. It provides a client for the ZenMux image generation API. The `scripts/generate.py` script correctly handles arguments, reads the `ZENMUX_API_KEY` from environment variables, encodes local images to base64 for API submission, and saves the generated image from the API response. There is no evidence of data exfiltration, malicious command execution, persistence mechanisms, or prompt injection attempts in `SKILL.md` to subvert the agent's intended behavior. All actions are directly aligned with the stated purpose of generating images via the ZenMux API.