← 返回 Skills 市场

Bee 视频全流程自动化

Name: Bee 视频全流程自动化
Author: jerryxn

作者 JerryXn · GitHub ↗ · v1.0.0 · MIT-0

cross-platform ⚠ suspicious

总下载

当前安装

版本数

在 OpenClaw 中安装

/install bee-video-workflow

功能描述

视频全流程自动化：下载 → 截封面 → OSS上传 → 蚁小二多平台分发 → 飞书多维表格记录。发一个视频链接就能跑完全程。

使用说明 (SKILL.md)

🐝 Bee — 视频全流程自动化

一条链接，全自动搞定：下载 → 封面 → OSS → 多平台发布 → 飞书记录。

环境变量

变量	说明	必填
`OSS_ACCESS_KEY_ID`	阿里云 AccessKey ID	✅
`OSS_ACCESS_KEY_SECRET`	阿里云 AccessKey Secret	✅
`OSS_BUCKET_NAME`	OSS Bucket 名称	✅
`OSS_ENDPOINT`	OSS Endpoint（如 oss-cn-hangzhou.aliyuncs.com）	✅
`OSS_PREFIX`	OSS 路径前缀（默认 `bee/`）
`YIXIAOER_TOKEN`	蚁小二 API Token（发布用）	发布时需要

飞书多维表格

默认配置在脚本头部：

App Token: Lv3fbmpWGacqhosSRtXct9EnnVd
Table ID: tblWsUQPqlAYsnV3

使用方式

自然语言（推荐）

直接跟 Agent 说：

"帮我处理这个视频：https://v.douyin.com/xxxxx"
"把这个视频发到所有平台"
"下载这个视频存到OSS"

命令行

# 完整流程：下载 → OSS → 蚁小二 → 飞书
python3 scripts/bee.py run "视频链接或本地路径" --title "标题" --publish 小桃犟 --tags '["标签1"]'

# 只下载 + OSS
python3 scripts/bee.py run "视频链接或本地路径" --title "标题"

# 只上传本地视频到 OSS + 记录
python3 scripts/bee.py run /path/to/video.mp4 --title "我的视频"

# 查看蚁小二账号
python3 scripts/bee.py accounts

# 查看发布状态
python3 scripts/bee.py status \x3CtaskSetId>

参数

参数	说明
`--title`	视频标题（不填则用原始标题）
`--desc`	视频描述
`--tags`	标签 JSON 数组，如 `'["搞笑","动画"]'`
`--publish`	蚁小二发布目标（账号名/分组名），不填则跳过发布
`--platform`	指定平台（抖音/小红书/视频号），不填则发到目标下所有平台
`--draft`	存草稿不发布
`--no-feishu`	跳过飞书记录
`--no-oss`	跳过 OSS 上传
`--cover`	自定义封面图路径（不填则自动从视频截取）

流程图

输入（链接/文件）
    │
    ▼
[1] 下载视频（支持抖音/本地文件/URL）
    │
    ▼
[2] 视频处理（ffprobe 获取信息 + ffmpeg 截封面）
    │
    ▼
[3] 上传到阿里云 OSS
    │
    ▼
[4] 蚁小二多平台发布（可选）
    │
    ▼
[5] 飞书多维表格记录
    │
    ▼
输出 JSON 结果

安全使用建议

This skill's high-level goal matches its code, but there are several red flags you should address before using it: - Do not supply your OSS credentials until you are comfortable: the script will upload files to OSS using the OSS_ACCESS_KEY_ID/SECRET you provide; grant only an account with minimal upload permissions. - Confirm and remove/replace the hardcoded Feishu app token if it isn't yours. A token in code may leak access or cause records to be written to someone else's table. - Ensure required runtime dependencies are present: install yt-dlp and the Python oss2 package, or update the skill metadata to declare them. Without them the skill will fail or behave unexpectedly. - The publish step relies on an external script at '../yixiaoer-pro/scripts/upload_and_publish.py' which is not included; inspect that script (or provide your own) before enabling publishing. If you don't have that script, publishing will be skipped but the skill will log that it tried. - The code will look for a cookies file at a parent-level path; make sure there are no sensitive cookies there you don't want read. Consider running the skill in an isolated environment or container. If you can't verify or fix the above (especially the hardcoded Feishu token and the missing external script), treat this skill as untrusted and avoid giving it your real OSS credentials. If you decide to proceed, run it in a sandbox, review/replace embedded credentials, and add explicit dependency documentation (yt-dlp, oss2).

功能分析

Type: OpenClaw Skill Name: bee-video-workflow Version: 1.0.0 The skill automates video downloading and publishing but contains hardcoded Feishu Bitable credentials (BITABLE_APP_TOKEN: Lv3fbmpWGacqhosSRtXct9EnnVd) in scripts/bee.py and SKILL.md, which causes video metadata (titles, authors, and OSS URLs) to be sent to a specific external table by default. Additionally, the script attempts to access files and execution scripts located multiple levels above its own directory (e.g., douyin_cookies.txt and yixiaoer-pro/scripts/upload_and_publish.py), which is a risky pattern that breaks bundle isolation and relies on a specific, potentially sensitive host environment.

能力评估

⚠ Purpose & Capability

Name/description match the code: downloading videos, extracting covers, uploading to OSS, calling a third-party publish tool, and recording to Feishu. However, the script uses yt-dlp and the Python oss2 library (for download and OSS upload) but the skill metadata only declares ffmpeg and ffprobe. The script also expects an external 'yixiaoer-pro' script at a filesystem path outside the skill bundle, which is not included or documented — this is disproportionate to the stated single-file automation purpose.

⚠ Instruction Scope

SKILL.md instructs the agent to run the included script, which is expected, but the runtime instructions and code reference additional resources not declared in the skill (yt-dlp binary, Python oss2 package, and a cookies file at a parent-level path). The script also emits an instruction for the agent to invoke an external feishu_bitable_create_record tool. Reading or using cookie files and invoking an external script located outside the skill expands the scope beyond what the description/metadata claim.

⚠ Install Mechanism

There is no install spec (instruction-only), which lowers install-time risk, but the runtime requires third-party tools and libraries that are not declared: the code calls 'yt-dlp' and imports 'oss2' (Python SDK) but neither is listed in required binaries/dependencies. This mismatch may cause runtime failures or hide additional implicit requirements that the user must install manually.

⚠ Credentials

The declared required env vars are for Alibaba OSS (OSS_ACCESS_KEY_ID, OSS_ACCESS_KEY_SECRET, OSS_BUCKET_NAME, OSS_ENDPOINT), which are appropriate for upload. However, the code contains a hardcoded Feishu app token and table ID (BITABLE_APP_TOKEN, BITABLE_TABLE_ID) baked into the script — a credential-like value embedded in code rather than declared or documented — and the code will look for a cookies file at a parent directory path if present. The skill also optionally uses YIXIAOER_TOKEN and YIXIAOER_BASE_URL for publishing, but the main metadata did not list YIXIAOER_TOKEN as required; overall, credential handling is inconsistent and partially opaque.

✓ Persistence & Privilege

The skill does not request always:true or any elevated persistent privileges. It writes to a local 'workspace' directory (creates workspace in the repo parent), which is normal for transient workflows. It does not modify other skills' configurations. The agent-invocation defaults are normal.

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 bee-video-workflow
安装完成后，直接呼叫该 Skill 的名称或使用 /bee-video-workflow 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

初始版本：视频下载→ffmpeg截封面→阿里云OSS上传→蚁小二多平台分发→飞书多维表格记录

元数据

Slug bee-video-workflow

版本 1.0.0

许可证 MIT-0

累计安装 0

当前安装数 0

历史版本数 1

常见问题