← 返回 Skills 市场
13605
总下载
22
收藏
0
当前安装
2
版本数
在 OpenClaw 中安装
/install agent-browser-core
功能描述
OpenClaw skill for the agent-browser CLI (Rust-based with Node.js fallback) enabling AI-friendly web automation with snapshots, refs, and structured commands.
安全使用建议
Install a trusted pinned version of the external CLI, run it in a dedicated low-privilege environment, allowlist target sites, keep saved login state minimal and protected, and require human approval before using eval, local file access, downloads, proxies, traffic interception, or credential/session changes.
功能分析
Type: OpenClaw Skill
Name: agent-browser-core
Version: 1.0.1
The skill bundle consists entirely of documentation and metadata for an 'agent-browser' CLI tool. All markdown files (SKILL.md and references/*.md) consistently emphasize security best practices, explicitly warn against high-risk capabilities like `eval`, `--allow-file-access`, and network manipulation, and instruct the AI agent to prioritize safe mode defaults and human approval for sensitive actions. There is no evidence of malicious intent, data exfiltration, unauthorized execution, or prompt injection attempts to subvert the agent; instead, the instructions guide the agent towards secure and controlled operation.
能力评估
Purpose & Capability
The skill enables AI-guided browser automation, including navigation, clicking, typing, snapshots, sessions, and authenticated workflows; these are sensitive capabilities but they fit the stated purpose and are disclosed.
Instruction Scope
The runtime guidance consistently labels high-risk features such as eval, local file access, traffic manipulation, cookies, storage, credentials, custom runtime controls, and downloads as needing task necessity or explicit approval.
Install Mechanism
No executable code is bundled; the skill points users to install an external agent-browser CLI and browser runtime, with version pinning, dedicated-environment use, and low-privilege execution guidance.
Credentials
The instructions recommend domain allowlists, blocking localhost and private network targets, avoiding elevated privileges, preferring read-only actions, and redacting secrets, which is proportionate for browser automation.
Persistence & Privilege
Background daemon use and browser state reuse are disclosed, and the skill advises ephemeral sessions where possible and treating saved state files as secrets.
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 agent-browser-core - 安装完成后，直接呼叫该 Skill 的名称或使用
/agent-browser-core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
v1.0.1
- Added comprehensive safety guidelines in a new `agent-browser-safety.md` reference file.
- Updated orientation and documentation links to include safety and high-risk operation controls.
- Introduced "Safe mode defaults" section to the main documentation, outlining recommended restrictions for secure automation.
- Revised supporting references to clarify safe usage practices.
v1.0.0
Initial release of agent-browser-core skill – enables AI-friendly web automation via agent-browser CLI.
- Supports Rust-based agent-browser with Node.js fallback for reliability.
- Provides structured commands, compact web snapshots, and ref-based automation.
- Includes guidance documents for installation, command usage, workflows, and troubleshooting.
- Emphasizes deterministic automation suited for AI agent workflows.
- Highlights security best practices and operational guardrails for web automation.
元数据
常见问题
Agent Browser Core 是什么？
OpenClaw skill for the agent-browser CLI (Rust-based with Node.js fallback) enabling AI-friendly web automation with snapshots, refs, and structured commands. 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件，目前累计下载 13605 次。
如何安装 Agent Browser Core？
在 OpenClaw 或 Claude Code 对话框中运行命令「/install agent-browser-core」即可一键安装，无需额外配置。
Agent Browser Core 是免费的吗？
是的，Agent Browser Core 完全免费（开源免费），可自由下载、安装和使用。
Agent Browser Core 支持哪些平台？
Agent Browser Core 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 Agent Browser Core？
由 codedao12（@codedao12）开发并维护，当前版本 v1.0.1。
推荐 Skills