← 返回 Skills 市场
234
总下载
0
收藏
1
当前安装
1
版本数
在 OpenClaw 中安装
/install taizi-agent-browser
功能描述
A fast Rust-based headless browser automation CLI with Node.js fallback that enables AI agents to navigate, click, type, and snapshot pages via structured co...
安全使用建议
This skill appears to be a wrapper for an external npm CLI (agent-browser). Before installing or using it, verify the upstream package and repository: 1) Look up the 'agent-browser' package on the npm registry and confirm the package name, maintainer, and recent versions match the skill's documentation. 2) Inspect the package contents (npm pack, unpack locally) or review the GitHub repository code to ensure there are no post-install scripts or downloads from unknown URLs, native addons, or unaudited binaries. 3) Avoid global installs on sensitive machines; test inside an isolated container or VM. 4) Because a headless browser can access arbitrary websites and persist cookies/storage, avoid running it with credentials or against sensitive sites until you're comfortable with the package provenance. 5) The mismatched metadata (different ownerId/slug/version between registry metadata and _meta.json) is suspicious — if you cannot reconcile the source and maintainer, prefer not to install. If you want help, provide the exact npm package page or repository URL and I can re-evaluate with that provenance.
功能分析
Type: OpenClaw Skill
Name: taizi-agent-browser
Version: 1.0.0
The skill provides a comprehensive wrapper for the 'agent-browser' CLI, granting the AI agent high-risk capabilities including arbitrary JavaScript execution via 'eval', access to session cookies and local storage, and network request interception/routing. While these features are aligned with the stated purpose of advanced browser automation, they represent a significant attack surface for data exfiltration or session hijacking if the agent is misdirected. The skill allows broad execution of any 'agent-browser' command through the Bash tool as defined in SKILL.md, including installation commands that may fetch external dependencies.
能力评估
Purpose & Capability
The SKILL.md documents a Node-based CLI for browser automation which reasonably requires node/npm. However there are several mismatches: the registry metadata names this skill 'taizi-agent-browser' while the included _meta.json and SKILL.md refer to 'agent-browser' and a different ownerId/version; source/homepage are marked unknown/none. The description also mentions a Rust binary with Node fallback but the instructions only show installing an npm package or building from a GitHub repo — this ambiguous provenance and metadata mismatch is unexpected for a well-scoped wrapper.
Instruction Scope
The SKILL.md confines runtime behavior to invoking the agent-browser CLI (navigate, snapshot, click, screenshot, etc.), which aligns with the stated purpose. It does instruct global npm installation and running 'agent-browser install' which may pull/execute additional artifacts. The allowed-tools entry (Bash(agent-browser:*)) means the agent will run that CLI directly; this is expected but gives the agent ability to execute arbitrary commands supported by that CLI. The instructions themselves do not ask to read unrelated local files or env vars.
Install Mechanism
There is no formal install spec — it's an instruction-only skill that tells users to run 'npm install -g agent-browser' or to clone a GitHub repo. That is a user action that will fetch and run third-party code outside the skill bundle. Because the package/package source is not verified here (no homepage, source unknown, and internal _meta.json disagrees with registry metadata), installing the npm package could introduce arbitrary native or JS code. The 'From Source' instructions point at a GitHub path but ownership/version inconsistencies reduce confidence.
Credentials
Declared requirements are limited to node and npm binaries; no environment variables, credentials, or config paths are requested by the skill itself. That is proportionate to a CLI wrapper that runs a Node-based tool.
Persistence & Privilege
always is false and the skill does not request persistent system-wide changes in its bundle. The skill allows the agent to invoke the CLI autonomously (disable-model-invocation=false) which is normal for skills. No indications the skill writes or modifies other skills or global agent settings were found.
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 taizi-agent-browser - 安装完成后，直接呼叫该 Skill 的名称或使用
/taizi-agent-browser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 taizi-agent-browser.
- Fast Rust-based headless browser automation CLI with Node.js fallback.
- Enables navigation, clicking, typing, and snapshotting via structured commands.
- Supports form filling, data extraction, UI testing, screenshots, and video recording.
- Provides semantic locators, JavaScript evaluation, cookie/storage management, and robust state handling.
- Includes quick installation and extensive examples for automating web workflows.
元数据
常见问题
Taizi Agent Browser 是什么？
A fast Rust-based headless browser automation CLI with Node.js fallback that enables AI agents to navigate, click, type, and snapshot pages via structured co... 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件，目前累计下载 234 次。
如何安装 Taizi Agent Browser？
在 OpenClaw 或 Claude Code 对话框中运行命令「/install taizi-agent-browser」即可一键安装，无需额外配置。
Taizi Agent Browser 是免费的吗？
是的，Taizi Agent Browser 完全免费，采用 MIT-0 许可证，可自由下载、安装和使用。
Taizi Agent Browser 支持哪些平台？
Taizi Agent Browser 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 Taizi Agent Browser？
由 fresh3（@fresh3）开发并维护，当前版本 v1.0.0。
推荐 Skills