← 返回 Skills 市场
111
总下载
0
收藏
1
当前安装
1
版本数
在 OpenClaw 中安装
/install opencli-browser
功能描述
Use when an agent needs to drive a real Chrome window via opencli — inspect a page, fill forms, click through logged-in flows, or extract data ad-hoc. Covers...
安全使用建议
Before installing or enabling this skill: 1) Confirm you have the opencli CLI and the required Chrome extension/debug-port configuration — the skill's metadata does not declare these but the runtime docs require them. 2) Treat this skill as powerful: it can read page DOM and network traffic (including session tokens or private form data). Avoid running it while logged into sensitive accounts; prefer a dedicated browser profile or test account. 3) Ask the skill author to update the manifest to declare required binaries/env vars and to document how sensitive data is handled. 4) If you must use it, restrict the agent's ability to invoke the skill autonomously or require user confirmation for each run.
功能分析
Type: OpenClaw Skill
Name: opencli-browser
Version: 1.0.0
The skill bundle provides comprehensive documentation and instructions for an AI agent to use the 'opencli' tool for browser automation. It covers element selection, form interaction, network traffic inspection, and session management with a focus on reliability and token efficiency. There are no indicators of malicious intent, data exfiltration, or prompt injection attacks; the content is entirely aligned with its stated purpose of enabling agentic browser control.
能力评估
Purpose & Capability
The SKILL.md repeatedly instructs the agent to run the opencli CLI and to interact with a running Chrome + extension (e.g., 'opencli doctor', 'opencli browser ...', references to Chrome debug port), but the registry metadata declares no required binaries, no install steps, and no required config paths. A legitimate opencli-based browser driver should at minimum declare opencli (and/or Chrome/Chromium) as a runtime dependency and document any required extension or debug-port config; the absence of those declarations is an incoherence.
Instruction Scope
The instructions themselves stay within the stated purpose (driving a live browser). They explicitly encourage inspecting page state, capturing network requests, running JS eval (read-only), and verifying writes. Those actions necessarily let the agent read page DOM, form contents, and network payloads (potentially including session identifiers or personal data). That is expected for a browser-driving skill, but it means the agent will have access to sensitive user data; the SKILL.md does not provide any additional constraints or warnings about that beyond usage best-practices.
Install Mechanism
This is an instruction-only skill with no install spec or code files. That minimizes disk-write risk. However, because it relies on an external CLI (opencli) and browser/extension setup, the manifest should have declared those runtime requirements.
Credentials
The SKILL.md references environment variables (OPENCLI_WINDOW_FOCUSED, OPENCLI_LIVE) and external browser state, but the declared requires.env is empty. More importantly, the skill's runtime actions (DOM reads, network capture) can expose highly sensitive data — cookies, auth tokens, page content — yet no explicit requirements/consent or warnings are declared in the metadata. The absence of declared binaries/envs and lack of an explicit statement about sensitive data handling is disproportionate to the manifest.
Persistence & Privilege
always:false and default autonomous invocation settings are used (user-invocable, agent may call autonomously). This is the platform default and not itself a problem. Note: because the skill lets the agent control a browser and capture network/page state, autonomous invocation increases risk; exercise caution if you allow the agent to call this skill without manual approval.
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 opencli-browser - 安装完成后，直接呼叫该 Skill 的名称或使用
/opencli-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
opencli-browser 1.0.0 – Initial release
- Enables structured control of a live Chrome window via opencli for agents.
- Provides selector-first target contract, compound form field support, and resilient element references.
- Handles stale reference recovery, structured error reporting, and network/API capture.
- Outputs machine-readable envelopes describing confidence and match quality for every action.
- Includes safety and robustness rules for agents: always inspect first, chain context-aware actions, and verify critical writes.
- Not intended for adapter development—use opencli-adapter-author for that workflow.
元数据
常见问题
opencli-browser 是什么？
Use when an agent needs to drive a real Chrome window via opencli — inspect a page, fill forms, click through logged-in flows, or extract data ad-hoc. Covers... 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件，目前累计下载 111 次。
如何安装 opencli-browser？
在 OpenClaw 或 Claude Code 对话框中运行命令「/install opencli-browser」即可一键安装，无需额外配置。
opencli-browser 是免费的吗？
是的，opencli-browser 完全免费，采用 MIT-0 许可证，可自由下载、安装和使用。
opencli-browser 支持哪些平台？
opencli-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 opencli-browser？
由 chang（@liberalchang）开发并维护，当前版本 v1.0.0。
推荐 Skills