Skill flagged — suspicious patterns detected

ClawHub Security flagged this skill as suspicious. Review the scan results before using.

Byted Las Video Inpaint

v1.0.1

Removes unwanted visual elements from videos using AI-powered inpainting via Volcengine LAS. Video watermark removal, subtitle removal, logo removal, and tex...

0· 126·0 current·0 all-time

Install

OpenClaw Prompt Flow

Install with OpenClaw

Best for remote or guided setup. Copy the exact prompt, then paste it into OpenClaw for volcengine-skills/byted-las-video-inpaint.

Previewing Install & Setup.
Prompt PreviewInstall & Setup
Install the skill "Byted Las Video Inpaint" (volcengine-skills/byted-las-video-inpaint) from ClawHub.
Skill page: https://clawhub.ai/volcengine-skills/byted-las-video-inpaint
Keep the work scoped to this skill only.
After install, inspect the skill metadata and help me finish setup.
Use only the metadata you can verify from ClawHub; do not invent missing requirements.
Ask before making any broader environment changes.

Command Line

CLI Commands

Use the direct CLI path if you want to install manually and keep every step visible.

OpenClaw CLI

Bare skill slug

openclaw skills install byted-las-video-inpaint

ClawHub CLI

Package manager switcher

npx clawhub@latest install byted-las-video-inpaint
Security Scan
Capability signals
Requires sensitive credentials
These labels describe what authority the skill may exercise. They are separate from suspicious or malicious moderation verdicts.
VirusTotalVirusTotal
Suspicious
View report →
OpenClawOpenClaw
Suspicious
medium confidence
Purpose & Capability
The skill claims to wrap Volcengine LAS inpainting (lasutil CLI) which fits its description. However the registry metadata lists no required environment variables while SKILL.md explicitly requires LAS_API_KEY (and references LAS_REGION and optionally VOLCENGINE_ACCESS_KEY/VOLCENGINE_SECRET). That metadata/instructions mismatch is inconsistent and should be corrected before trusting automated installation.
!
Instruction Scope
The SKILL.md instructs the agent to source scripts/env_init.sh (which performs network fetches and pip installs) and to create/consume a local .env or env.sh containing LAS_API_KEY. It also directs uploading local files to TOS and requires the user to provide output_tos_path; SKILL.md asks for extra credentials (VOLCENGINE_ACCESS_KEY/SECRET) in some flows. Asking users to place secrets in a local file and sourcing it expands scope (reads local files/env) beyond a pure 'instruction-only' wrapper and could expose credentials if mishandled.
!
Install Mechanism
There is no declared install spec, but scripts/env_init.sh performs remote operations: curl a manifest from las-ai-cn-beijing-online.tos-cn-beijing.volces.com and pip-install a wheel directly from that host. Pulling and installing a remote wheel at runtime is effectively executing remote code without an explicit install declaration in the registry—this is a meaningful risk and should be treated as an install step requiring review.
!
Credentials
The skill needs LAS_API_KEY (declared only in SKILL.md, not in registry metadata). It also mentions LAS_REGION and, in some flows, VOLCENGINE_ACCESS_KEY/VOLCENGINE_SECRET for downloading outputs. Those additional credentials are not declared in the registry and may grant broader access to a user's cloud storage. The number and type of credentials requested should be explicitly declared and minimized.
Persistence & Privilege
always:false (good). However env_init.sh will create/activate a virtualenv (.las_venv) in the project and pip-install a remote wheel, and scripts create temporary workdirs under /tmp. These actions change the local environment and write files; they're expected for a CLI wrapper but worth noting because the script performs network fetch-and-install at runtime.
What to consider before installing
Key things to consider before installing or running this skill: - Metadata mismatch: SKILL.md requires LAS_API_KEY (and mentions LAS_REGION and possibly VOLCENGINE_ACCESS_KEY/SECRET) but the registry metadata lists no required env vars. Treat those credentials as required and confirm you are comfortable providing them. - Remote code fetch: env_init.sh curls a manifest and pip-installs a wheel from las-ai-cn-beijing-online.tos-cn-beijing.volces.com at runtime. That effectively executes remote code—do not run this in a sensitive environment without audit. Prefer downloading the wheel manually, inspecting it, or running inside an isolated sandbox/VM. - Secrets handling: The skill suggests creating a local env file with your API key and sourcing it. Never paste secrets into chat. If you must provide keys, use least-privileged credentials and consider temporary keys with limited scope and lifetime. - Output location and access: The skill requires a user-owned TOS output path (tos://bucket/...). Confirm you control that bucket and credentials before submitting jobs. If the skill later asks for VOLCENGINE_ACCESS_KEY/SECRET to download results, limit those keys to the specific bucket and revoke when finished. - Verification steps: Ask the publisher for provenance (official Volcengine source, homepage). If you can't verify, run the skill only in an isolated environment, inspect scripts (especially env_init.sh and the pip wheel) before permitting network installation, and ensure you get an explicit list of all environment variables the skill needs. Given the remote install behavior and metadata inconsistencies, treat this skill as suspicious until you can validate the downloaded wheel and confirm the credential requirements.

Like a lobster shell, security has layers — review code before you run it.

latestvk9728j9rbgw895qngncwg8ybdd851ktx
126downloads
0stars
2versions
Updated 1w ago
v1.0.1
MIT-0

LAS 视频修复(las_video_inpaint

去除视频中的水印、字幕、Logo。支持指定固定区域修复(fixed_bboxes)。多种修复后端可选。

使用限制:视频时长 ≤ 4 小时,文件大小 ≤ 30GB。

设计模式

本 skill 主要采用:

  • Tool Wrapper:封装 lasutil CLI 调用
  • Pipeline:包含 Step 0 → Step N 的顺序工作流

核心 API 与配置

  • 算子 ID: las_video_inpaint
  • API: 异步(submitpoll
  • 环境变量: LAS_API_KEY (必填)

详细参数与接口定义见 references/api.md

Gotchas

  • 密钥安全:若聊天框屏蔽密钥,让用户在当前目录创建 env.sh 并写入 export LAS_API_KEY="...",SDK 会自动读取。
  • 免责声明:最终回复结果时必须包含:"本方式的计费均为预估计费,与实际费用有差距,实际费用以运行后火山产生的账单为准。计费说明请参考 Volcengine LAS 定价。",且禁止使用"实际费用"字眼描述预估价。

工作流(严格按步骤执行)

复制此清单并跟踪进度:

执行进度:
- [ ] Step 0: 前置检查
- [ ] Step 1: 初始化与准备
- [ ] Step 2: 预估价格
- [ ] Step 3: 提交任务
- [ ] Step 4: 异步查询
- [ ] Step 5: 结果呈现

Step 0: 前置检查(⚠️ 必须在第一轮对话中完成)

在接受用户的任务后,不要立即开始执行,必须首先进行以下环境检查:

  1. 检查 LAS_API_KEYLAS_REGION:确认环境变量或 .env 中是否已配置。
    • 若无,必须立即向用户索要(提示:LAS_REGION 常见为 cn-beijing)。
    • 注意LAS_REGION 必须与您的 API Key 及 TOS Bucket 所在的地域完全一致。如果用户中途切换了 Region,必须提醒用户其 TOS Bucket 也需对应更换,否则会导致权限异常或上传失败。
  2. 检查输入路径
    • 如果用户要求处理的是本地文件,则需要先通过 File API 上传至 TOS(只需 LAS_API_KEY,无需额外 TOS 凭证)。
    • 如果算子的输出结果存放在 TOS 上,且用户需要下载回本地,则需要 VOLCENGINE_ACCESS_KEYVOLCENGINE_SECRET_KEY。对于仅需要上传输入文件的场景,TOS 凭证不再必须
  3. 检查输出路径
    • output_tos_path 为必填参数,必须由用户提供自己可写的 TOS 目录路径(格式:tos://bucket/output_dir/)。
    • 服务端需要将修复后的视频写入此目录。
  4. 确认无误后:才能进入下一步。

Step 1: 初始化与准备

环境初始化(Agent 必做)

# 执行统一的环境初始化与更新脚本(会自动创建/激活虚拟环境,并检查更新)
source "$(dirname "$0")/scripts/env_init.sh" las_video_inpaint
workdir=$LAS_WORKDIR

如果网络问题导致更新失败,脚本会跳过检查,使用本地已安装的 SDK 继续执行。

  • 处理本地文件时:使用 File API 上传(只需 LAS_API_KEY,无需 TOS 凭证和 Bucket):
    lasutil file-upload <local_path>
    
    上传成功后返回 JSON,取其中的 presigned_url(HTTPS 预签名下载链接,24 小时有效)传给算子作为输入 URL。

Step 2: 预估价格(⚠️ 必须获得用户确认)

  1. 读取 references/prices.md 获取最新计费标准。
  2. 获取视频时长:
    lasutil media-duration <video_url>
    
  3. 根据时长和模式单价计算总价,将计费单价与预估总价一并告知用户并强制暂停执行,明确等待用户回复确认。在用户明确回复"继续"、"确认"等同意指令前,绝对禁止进入下一步(执行/提交任务)。提示:预估仅供参考,实际以火山账单为准。计费说明请参考 Volcengine LAS 定价

Step 3: 提交任务 (Submit)

构造基础 data.json

{
  "video_url": "<presigned_url>",
  "output_tos_path": "tos://<your-bucket>/output_dir/",
  "targets": ["watermark", "subtitle"]
}

重要提示: output_tos_path 必须由用户提供,需要填写用户自己账号下可写的 TOS 目录(服务端会将修复后的视频写入此目录)。

单文件提交

data=$(cat "$workdir/data.json")
lasutil submit las_video_inpaint "$data" > "$workdir/submit.json"

⚠️ 强制反馈:任务提交成功后,必须立即向用户返回生成的 task_id,以便用户跟踪进度或在必要时手动查询。

Step 4: 异步查询 (Poll)

⚠️ 异步任务与后台轮询约束

  • 如果你当前的环境支持后台任务/异步长效运行:你可以利用环境提供的后台能力(例如发起后台轮询任务),并在任务完成后主动将结果返回给用户。
  • 如果你当前的环境不支持长效后台任务(如普通的单轮对话沙箱),且直接 sleep 循环会导致超时崩溃:绝对禁止在代码中执行死循环等待! 此时必须立即向用户输出 Task ID 并结束当前轮次,告知用户:"任务已提交,请稍后向我询问进度"。

单任务查询

lasutil poll las_video_inpaint {task_id}
  • COMPLETED → 返回修复后视频路径 result.data.inpainted_video_path
  • RUNNING/PENDING → 稍后重试。

Step 5: 结果呈现

处理结果

# 获取修复后的视频 URL
inpainted_url=$(cat "./output/{task_id}/result.json" | jq -r '.data.inpainted_video_path')
echo "修复后的视频: $inpainted_url"

视频文件

  • 修复后的视频已保存在 TOS,直接返回预签名 URL
  • 无需再次上传

向用户展示

  1. 修复后的视频下载链接
  2. 本地结果路径:./output/{task_id}/
  3. 计费声明

审查标准

执行完成后,Agent 应自检:

  1. 环境变量是否正确配置
  2. 输入文件是否成功上传
  3. 输出结果是否正确呈现给用户
  4. 计费声明是否包含

Comments

Loading comments...