内容引擎

内容引擎（小红书）。两种 mode：①拆解（v1）— 输入 XHS 爆款链接，输出 18 维结构化拆解卡；②生成（v2）— 在拆解卡基础上结合品牌信息，用 Ofox（LLM + Nano Banana）生成我方版本的脚本/文案/素材图/封面/标签全套产出。视频生成（Seedance 2.0）规划在 v2.1。同时...

Key things to check before installing or running this skill: 1) Missing declared credentials: The code expects at least TIKHUB_API_TOKEN and likely credentials for Ofox (LLM), Nano Banana (image generation) and Seedance (video). The skill metadata lists no required env vars — ask the author (or inspect llm.py/nano_banana.py/seedance.py) to get the exact env var names and endpoints. 2) .env lookup behavior: The client will search for a .env in your current working directory and in your XDG config directory (~/.config/content-engine/.env). If you keep sensitive tokens in those locations for other projects, the skill might read and use them. Run in an isolated directory or remove/rename unrelated .env files before use. 3) Data exfiltration surface: The pipeline downloads note content, video, frames and comments and sends them to external services (LLM/image/video providers). Confirm which external hosts are used and whether you are comfortable sending that data (privacy, IP, compliance). Consider using throwaway/dev keys and limits when first testing. 4) Filesystem changes: The skill will write outputs and also auto-append to graph/ (it modifies files inside the skill). If you need the skill to be read-only, avoid running modes that auto-write or run it from a copy. 5) Run preflight and audit code: Use the built-in --check / preflight mode to see which dependencies and tokens are required. Inspect llm.py, nano_banana.py, seedance.py for exact network endpoints and credential names. Prefer running the scripts in a sandboxed environment or container with restricted network access until you verify behavior. 6) Operational advice: Provide only dedicated, least-privilege API keys (not long-lived platform/root credentials). If you must use production tokens, restrict them (scopes, IP allowlist) and monitor usage. If anything is unclear from the author, treat the mismatch between code and declared requirements as a red flag and avoid giving sensitive keys to this skill until resolved.

Type: OpenClaw Skill Name: qianxun-content-engine Version: 0.3.1 The qianxun-content-engine skill bundle is a legitimate and well-documented tool designed for analyzing and generating social media content for the Xiaohongshu (XHS) platform. It utilizes Python standard libraries and the ffmpeg binary to perform tasks such as metadata extraction, comment parsing, and video frame processing. The code integrates with external APIs (TikHub, Ofox, and Volcengine Ark) using standard authentication patterns and includes security best practices like file size limits and URL validation to prevent DoS and SSRF. The instructions in SKILL.md are strictly task-aligned, focusing on content deconstruction and generation workflows without any evidence of malicious intent, data exfiltration, or unauthorized persistence.