← 返回 Skills 市场
数字人口播生成
作者
XuHongFeii2
· GitHub ↗
· v1.0.0
· MIT-0
95
总下载
0
收藏
0
当前安装
1
版本数
在 OpenClaw 中安装
/install shuziren-koubo-shengcheng
功能描述
可以生成数字人口播视频。训练自己的数字人、生成口播视频。
安全使用建议
This skill implements the advertised workflow but you should do the following before enabling it: 1) Verify you trust the platform domain (http://easyclaw.bar) because the skill will send whatever API token/keys you provide to that host; 2) Confirm the registry metadata is updated to list CHANJING_PLATFORM_API_TOKEN (or API_KEY/API_SECRET) as required — treat the missing declaration as a red flag until clarified; 3) Review the included scripts yourself (they are provided) to confirm their behavior; 4) Be aware the skill will attempt to create recurring OpenClaw cron jobs and write state to ~/.openclaw/cron/watchers/ — if you do not want persistent background jobs, do not configure the token or run the watcher commands; 5) If you decide to proceed, prefer creating and supplying a scoped/platform-issued token (not your upstream Chanjing app_id/secret) and avoid sharing high-privilege credentials until you've validated the platform.
功能分析
Type: OpenClaw Skill
Name: shuziren-koubo-shengcheng
Version: 1.0.0
This skill bundle provides an integration for a digital human video generation service via a platform API (easyclaw.bar). It includes scripts for listing resources, training assets, and synthesizing videos, utilizing OpenClaw's native cron system (`schedule_task_watch.py`) to monitor long-running asynchronous tasks. The implementation follows security best practices such as using `shlex.quote` for shell command construction and providing explicit instructions in `SKILL.md` to guide the agent's behavior and prevent unauthorized API probing. All high-privilege actions, including network requests and local script execution, are directly aligned with the stated purpose of the skill.
能力评估
Purpose & Capability
Name/description (generate digital-human talking videos, train digital humans/voices) matches the included Python scripts and platform API calls. Requiring Python and making POST/GET calls to /openclaw/* endpoints is coherent. However the registry metadata declares no required environment variables or primary credential while the SKILL.md and client code clearly expect a platform token or legacy key/secret — an inconsistency between metadata and runtime requirements.
Instruction Scope
The SKILL.md instructs the agent to only run the provided scripts (good) and avoid ad-hoc HTTP calls. The scripts do network I/O to a specific platform (easyclaw.bar/shuzirenapi), upload user-specified files, write watcher state under ~/.openclaw/cron/watchers/, and invoke the local OpenClaw CLI to schedule cron jobs. Those actions are within the skill's stated purpose (scheduling watchers for asynchronous tasks), but they introduce side effects on the host (file writes, subprocess execution, scheduled jobs) that a user should expect and approve.
Install Mechanism
No install spec is present (scripts are included and run directly), which is low-risk from an installer perspective. There is no external archive download or package install that would write arbitrary code during installation.
Credentials
The skill expects CHANJING_PLATFORM_API_TOKEN (preferred) or CHANJING_PLATFORM_API_KEY / CHANJING_PLATFORM_API_SECRET as fallbacks, and may read CHANJING_PLATFORM_BASE_URL. These are reasonable for talking to the platform, but the registry metadata did not declare them as required or a primary credential — a metadata omission. The client will send whatever token/keys are present to easyclaw.bar; ensure you trust that domain before supplying secrets. The scripts also consult other environment variables (OPENCLAW_BIN, OPENCLAW_INSTALL_DIR, LOCALAPPDATA) to find the OpenClaw CLI — these are not declared but are used only to locate the launcher.
Persistence & Privilege
The skill schedules background watcher jobs via the local OpenClaw CLI (cron create / cron add) and writes state files under the user's home directory. While this supports the advertised asynchronous-watch behavior, it effectively grants the skill the ability to create recurring local commands that will execute later (including passing the platform token). This is significant persistence and should be acceptable only if you trust the platform endpoint and the exact commands being scheduled.
如何使用
- 确保已安装 OpenClaw(本地或 Docker 部署)
- 在对话框中输入安装命令:
/install shuziren-koubo-shengcheng - 安装完成后,直接呼叫该 Skill 的名称或使用
/shuziren-koubo-shengcheng触发 - 根据 Skill 的参数说明提供必要输入,即可获得结构化输出
版本历史
v1.0.0
数字人口播生成 Skill v1.0.0
- 新增用于生成数字人口播视频的能力,包括训练专属数字人、生成口播视频等。
- 明确所有操作需通过 scripts/ 目录下的入口 Python 脚本实现,不可手写 HTTP 请求。
- 强化异步任务流程,集成 OpenClaw cron watcher,实现生成完成自动通知。
- 丰富口播视频配置项,支持字幕、背景、分辨率等多种可选参数。
- 明确环境变量和 Token 获取指引,说明接口优先事项和防止重复任务提交的规则。
- 统一响应格式,要求输出图片/视频直链并优先渲染相关媒体内容。
元数据
常见问题
数字人口播生成 是什么?
可以生成数字人口播视频。训练自己的数字人、生成口播视频。 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件,目前累计下载 95 次。
如何安装 数字人口播生成?
在 OpenClaw 或 Claude Code 对话框中运行命令「/install shuziren-koubo-shengcheng」即可一键安装,无需额外配置。
数字人口播生成 是免费的吗?
是的,数字人口播生成 完全免费,采用 MIT-0 许可证,可自由下载、安装和使用。
数字人口播生成 支持哪些平台?
数字人口播生成 跨平台运行,可在任意部署了 OpenClaw / Claude Code 的环境中使用(cross-platform)。
谁开发了 数字人口播生成?
由 XuHongFeii2(@xuhongfeii2)开发并维护,当前版本 v1.0.0。
推荐 Skills