Pixcli Skill

Creative toolkit for AI agents — generate images, videos, voiceover, music, and sound effects, then assemble polished output via Remotion. Uses the pixcli CL...

This skill appears to be what it claims: a Node CLI that calls a pixcli service and provides Remotion templates. Before installing or running it: - Treat the PIXCLI_API_KEY like any API secret: create a limited key if possible and avoid reusing high-privilege credentials. - Installing via npm (npm install -g pixcli) will pull code from the public registry — review the npm package and its maintainers (and choose npx for one-off use if you prefer not to install globally). - The README mentions an OPENROUTER_API_KEY fallback (not listed in the skill metadata). If you have that env var set, the CLI may use it — consider removing unrelated secrets from your environment when running third-party CLIs. - The Remotion templates included in the bundle will require npm install in a copied folder and then running npx remotion (which executes JS code locally). The templates in this bundle look benign, but review any third-party template code before rendering in a sensitive environment. - If you need higher assurance, inspect the upstream npm package source (pixcli on npm / its GitHub) and the runtime behavior of the CLI (particularly network endpoints and what metadata it sends) before providing any private or sensitive data.

Type: OpenClaw Skill Name: pixcli-skill Version: 2.2.0 The bundle is a comprehensive toolkit for AI-driven video production using the `pixcli` CLI and the Remotion framework. It is classified as suspicious because it grants the AI agent broad and high-risk permissions in `SKILL.md`, specifically the ability to execute arbitrary commands via `Bash(npm *)`, `Bash(node *)`, and `Bash(npx *)`. While these capabilities are plausibly required for managing Node.js-based video projects and installing dependencies, they represent a significant attack surface for shell injection or unauthorized code execution. No evidence of intentional malice, data exfiltration, or malicious prompt injection was found in the CLI logic, template code, or extensive documentation (IOC: pixcli.shellbot.sh).