← 返回 Skills 市场
138
总下载
0
收藏
0
当前安装
1
版本数
在 OpenClaw 中安装
/install visual-benchmarker
功能描述
(元技能) 视觉对标视频搜索器，通过指导AI调用其他工具，为项目确认视觉风格。
安全使用建议
This meta-skill is instruction-only and appears coherent: it extracts keywords from a user-provided text file and delegates video search to an installed douyin-video-search skill. Before installing/using it, confirm that you have (or trust) the douyin-video-search skill referenced in SKILL.md and that its tikhub_api_token (mentioned in the doc) is configured securely — the meta-skill will rely on that token when invoking the tool. Also avoid uploading sensitive secrets in the input text (the skill processes user-provided text), and if you want stronger assurance, request or inspect the douyin-video-search skill's manifest/source to see what network calls and credentials it actually uses.
功能分析
Type: OpenClaw Skill
Name: visual-benchmarker
Version: 1.0.0
The 'visual-benchmarker' skill is a purely instructional 'meta-skill' (SKILL.md) that guides an AI agent through a workflow of keyword extraction and video searching. It contains no executable code, does not request sensitive permissions, and relies on a legitimate dependency (douyin-video-search) to perform its stated function without any signs of malicious intent or prompt injection attacks.
能力评估
Purpose & Capability
The skill is explicitly a meta-skill whose job is to extract search keywords and orchestrate a call to an existing douyin-video-search tool to find example videos. The declared functionality matches the instructions and there are no unexpected binaries or credentials requested by this skill itself.
Instruction Scope
SKILL.md instructs the agent to read a user-supplied text file, produce search keywords, call the installed tools.douyin_video_search(...) interface, filter results and produce a Markdown report. The instructions are scoped to the stated purpose and explicitly forbid reading hard-coded local files. Note: the doc also says the douyin-video-search skill must be configured with a tikhub_api_token in config.json — the token requirement is for the dependent skill, not this meta-skill, but users should be aware the agent will rely on that configured credential when the tool is invoked.
Install Mechanism
No install spec and no code files — instruction-only means nothing is downloaded or written by this skill itself, which is the lowest-risk installation model.
Credentials
This skill declares no required environment variables or credentials itself. However, SKILL.md documents a dependency on an installed skill that requires tikhub_api_token in config.json. The meta-skill not declaring that dependency in registry metadata is a minor transparency gap: the meta-skill will effectively cause the agent to use a credential belonging to another configured skill.
Persistence & Privilege
Flags show always:false and default autonomous invocation behavior; the skill does not request elevated persistence or system-wide config changes and does not indicate it will modify other skills' settings.
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 visual-benchmarker - 安装完成后，直接呼叫该 Skill 的名称或使用
/visual-benchmarker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
Initial release of visual-benchmarker (元技能) – a workflow to help projects identify visual styles by orchestrating AI and external tools.
- Defines a 3-step process: extract search keywords, fetch Douyin videos using douyin-video-search skill, and select benchmark cases.
- Provides prompts and templates for AI to extract core categories and search terms from user documents.
- Specifies use of the douyin-video-search tool, with sample calls and usage guidelines.
- Outputs a Markdown report with at least 3 recommended videos, each with links and visual analysis.
- Emphasizes tool-based dynamic calls and quality of keyword extraction.
元数据
常见问题
Visual Benchmarker 是什么？
(元技能) 视觉对标视频搜索器，通过指导AI调用其他工具，为项目确认视觉风格。 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件，目前累计下载 138 次。
如何安装 Visual Benchmarker？
在 OpenClaw 或 Claude Code 对话框中运行命令「/install visual-benchmarker」即可一键安装，无需额外配置。
Visual Benchmarker 是免费的吗？
是的，Visual Benchmarker 完全免费，采用 MIT-0 许可证，可自由下载、安装和使用。
Visual Benchmarker 支持哪些平台？
Visual Benchmarker 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 Visual Benchmarker？
由 an（@ahsbnb）开发并维护，当前版本 v1.0.0。
推荐 Skills