Siluzan Cso

当用户提问的内容涉及以下内容时，可以使用本SKILL（1）多媒体平台内容(视频/图文)发布与运营（YouTube、TikTok、Instagram、LinkedIn、X、视频号），以及账号授权、数据报表、任务管理；（2）公众号、小红书等内容文案/选题生成——选题/拆解/口播成稿、三库选题；（3）RAG 知识库检索...

This skill's purpose and instructions are coherent for a social-media publishing/RAG content workflow, but there are a few things to double-check before installing: - Inspect the installer script before running it. SKILL.md asks you to run a remote installer from https://unpkg.com/siluzan-cso-cli@latest/dist/skill/scripts/install.sh (or the PowerShell equivalent). Open that URL (or fetch the package tarball from npm) and review the script contents and what it installs — do not run curl | bash / iex blindly. - Prefer installing the CLI from a trusted source manually (npm install -g siluzan-cso-cli) after reviewing the package page, sources, and maintainers. If possible, install in an isolated environment (container or VM) first. - The skill will read/write ~/.siluzan/config.json and can write files into agent skill directories (init -d). Only allow it if you trust the package and understand which paths will be modified. Avoid using the config set --token pattern that writes tokens into shell history; use environment variables (SILUZAN_API_KEY / SILUZAN_AUTH_TOKEN) or the interactive login flow the SKILL.md recommends. - Resolve the small metadata mismatch: _meta.json claims node/npm required while the top-level 'Requirements' showed none — ask the publisher to clarify required binaries and exactly what the install script does. - If you need higher assurance, ask the publisher for the install script contents, a pinned release tarball, or to review the source repo (the registry metadata points to an Azure DevOps repo and https://www.siluzan.com homepage). Verify signatures or checksums where available. If you cannot inspect the installer or trust the publisher, run the CLI only in an isolated environment or decline installation.

Type: OpenClaw Skill Name: siluzan-cso Version: 1.1.12 The skill bundle is a legitimate toolset for social media management and AI-driven content generation. It demonstrates high security awareness by explicitly forbidding the upload of sensitive directories (e.g., ~/.ssh, ~/.aws, .env) in references/upload.md and warning users against insecure credential handling in references/setup.md. While the installation uses a curl-to-shell pattern, it is standard for CLI tools and points to a public npm-based CDN (unpkg.com), and the AI instructions (SKILL.md) prioritize user confirmation for all write operations.