面向 faceless/no-face 场景的 Sparki skill 变体，沿用最新版官方 Sparki 安装、API key、上传和命令说明，同时保留 faceless 场景定位。

This skill appears to be a legitimate Sparki CLI variant tailored for 'faceless' videos, and it mostly does what it says: uploads local videos, creates edit projects, polls status, and downloads results. Before installing or running it, consider the following: - Verify the 'uv' binary: the install step runs 'uv sync' — confirm where 'uv' comes from and what 'uv sync' will do (it may run arbitrary sync/install steps). If you don't trust 'uv', don't run the install command. - Confirm network expectations: the manifest only lists agent-api.sparki.io, but the skill will download result files from URLs provided by the Sparki service (likely CDNs or S3). If your environment restricts outbound domains, update the manifest or allow the expected CDN domains. Be cautious if result URLs point to unknown hosts. - Environment variables: the code honors SPARKI_API_KEY (declared) and may use SPARKI_UPLOAD_TG_LINK if present (not declared). Only provide an API key you trust for the Sparki service; do not reuse high-privilege keys that grant broader access elsewhere. - If you need higher assurance, run the included Python code in an isolated environment (container or VM), review the uv tool source/manifest, and verify network traffic (where uploads/downloads go) before giving it production credentials. Given the mismatches above (network scope, undeclared env use, opaque installer), proceed with caution — these look like engineering sloppiness but could also enable unintended network activity if left unreviewed.

Type: OpenClaw Skill Name: faceless-video-zh Version: 1.0.12 The skill is a legitimate CLI wrapper for the Sparki AI video editing service, facilitating video uploads, automated editing, and downloads via the 'agent-api.sparki.io' endpoint. The code in 'src/sparki_cli/cli.py' and 'src/sparki_cli/client.py' follows standard API integration patterns using 'httpx' and 'typer', with no evidence of malicious execution, data exfiltration, or obfuscation. The instructions in 'SKILL.md' are functional directives intended to guide the AI agent's tool selection for video processing tasks rather than subverting its safety constraints.