Venice AI Media

Generate, edit, and upscale images; create videos from images via Venice AI. Supports text-to-image, image-to-video (Sora, WAN), upscaling, and AI editing.

This skill appears to do what it says: a Python-based Venice.ai client needing only a Venice API key. Before installing: (1) decide whether you are comfortable storing your VENICE_API_KEY on disk (the README suggests adding it to ~/.clawdbot/clawdbot.json); (2) expect the scripts to fetch remote images if you use --url and to possibly perform web searches if you use --enable-web-search — these flags cause extra network activity beyond Venice.ai API calls; (3) the scripts may auto-install Python packages (requests) at runtime, which will write to disk and contact package registries; (4) if you do not want the AI model to call this skill autonomously (which would allow the agent to use your VENICE_API_KEY without an explicit user action), disable model invocation for the skill. If you want extra assurance, review the provided Python scripts (scripts/*.py) to confirm exactly what network endpoints are contacted and how your API key is used.

Type: OpenClaw Skill Name: venice-ai-media Version: 1.0.2 The OpenClaw AgentSkills skill bundle is benign. All Python scripts (`venice-edit.py`, `venice-image.py`, `venice-upscale.py`, `venice-video.py`, `venice_common.py`) and the `SKILL.md` documentation are clearly aligned with the stated purpose of interacting with the Venice AI API for media generation and editing. The scripts use standard Python libraries for API communication (to `https://api.venice.ai`), read the `VENICE_API_KEY` from expected locations (environment or `~/.clawdbot/clawdbot.json`), and handle local file inputs/outputs as part of their core functionality. There is no evidence of data exfiltration to unauthorized endpoints, malicious execution, persistence mechanisms, obfuscation, or prompt injection attempts against the AI agent in the `SKILL.md`.