Browser Automation
/install bridgic-browser
Dependencies
A bridgic-browser project requires the following packages:
| Package | Description |
|---|---|
bridgic-browser |
Browser automation CLI + Python SDK (installing one installs both) |
Additionally, browser binaries must be installed once: uv run playwright install chromium.
Installation: Run the install script to set up all dependencies:
bash "skills/bridgic-browser/scripts/install-deps.sh" "$PWD"
The script checks uv availability, initializes a uv project if needed, installs missing packages, and ensures Playwright chromium is available.
Strategies & Guidelines (Important!!)
Notes:
- Whenever invoking the
bridgic-browserCLI, you must call it usinguv run. - If the user clearly specifies exact steps that must be followed, try to perform the exploration according to those steps. If loops or branches appear during exploration, decide the best exploration path autonomously.
- If you think you may need to return to the original page after clicking into a new page, try opening the new page in a new browser tab instead of using a “click then go back” approach. This is especially important when the original page already has interaction state (such as filled forms or applied filters); otherwise, that state may be lost after navigating back. Be sure to close the new tab promptly after finishing the related actions.
- If exploration involves repeatedly clicking items in a list, you do not need to traverse every item (especially when the list is large).
- If login, verification, or authorization is required during exploration, pause and ask the user to complete it manually, unless the user explicitly provides instructions in the task.
- To avoid operating on websites too frequently, maintain human-like access intervals during both exploration and coding. You may simulate random wait times to reduce the risk of being blocked. Note: the
bridgic-browser waitcommand parameter is in seconds, not milliseconds; for example,bridgic-browser wait 2orbridgic-browser wait 3.2. - After finishing exploration and code writing, automatically run testing/validation.
- CDP mode tab visibility: when attached via
--cdpto a user's running Chrome,tabs/switch-tab/close-tabonly see pages bridgic itself opened (the initial blank tab plus anything spawned from it vianew-tabor a click on atarget="_blank"link). The user's other tabs are deliberately invisible to bridgic — never assume you canswitch-tabinto them. To work with such a tab, ask the user to navigate to it through bridgic, or usenew-tab \x3Curl>.
Reference Files
Reference files cover all use cases. Load only the one(s) relevant to the task:
| Scenario | Interface | Load |
|---|---|---|
| Directly control browser from terminal | CLI | cli-guide.md |
| Write Python code about browser automation | Python | sdk-guide.md |
| Write shell script about browser automation | CLI | cli-guide.md |
| Explore via CLI, then generate Python code | CLI → Python | cli-sdk-api-mapping.md + sdk-guide.md |
| Migrate / compare / explain CLI ↔ SDK | Both | cli-sdk-api-mapping.md |
| Configure env vars or login state persistence | Either | env-vars.md |
Connect to an existing Chrome (chrome://inspect, --remote-debugging-port, cloud browser, Electron) |
CLI / SDK | cdp-mode.md |
Interface Decision Rules
- Output requested as shell commands or scripts → use CLI guide first (
references/cli-guide.md). - Output requested as runnable Python code (
async,Browser, tool builder) → use SDK guide first (references/sdk-guide.md). - Input is CLI outputs or actions but output needs to be Python code → use mapping guide first (
references/cli-sdk-api-mapping.md), then SDK guide for final code generation (references/sdk-guide.md). - If intent is ambiguous, infer from requested artifacts (
.sh/ terminal session vs.pyscript).
Common Usage (CLI + SDK)
- Ref-based actions depend on the latest snapshot.
- After navigation or major DOM updates, refs can become stale; refresh snapshot before ref actions.
- CLI keeps state in a daemon session across invocations. Set
BRIDGIC_HOMEenv var to run multiple independent daemon instances (each with its own socket, logs, and user data). - SDK keeps state in the Python process/context. By default, browser profile (cookies, session) is persisted to
$BRIDGIC_HOME/bridgic-browser/user_data/(default~/.bridgic/...); passclear_user_data=TruetoBrowser()for an ephemeral session. - Use exact command/method names from references; do not invent aliases.
Bridge Workflow: CLI Actions -> Python Code
- Parse CLI steps in order.
- Map each step using
references/cli-sdk-api-mapping.md. - Preserve behavior details: refs, options, arguments, configuration, etc.
- Emit runnable async Python code with explicit browser lifecycle (
async with Browser(...)preferred). - Call out any behavior differences that cannot be represented 1:1.
Minimal Quality Checklist
- CLI request: return valid CLI commands/options only.
- SDK request: return executable async Python with correct imports.
- Bridge request: include mapping rationale plus final SDK code.
- 确保已安装 OpenClaw(本地或 Docker 部署)
- 在对话框中输入安装命令:
/install bridgic-browser - 安装完成后,直接呼叫该 Skill 的名称或使用
/bridgic-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
Browser Automation 是什么?
Use for any task requiring a real browser: viewing web pages, accessing login-gated sites, operating web UIs, scraping social media (Xiaohongshu/Weibo/Twitte... 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件,目前累计下载 101 次。
如何安装 Browser Automation?
在 OpenClaw 或 Claude Code 对话框中运行命令「/install bridgic-browser」即可一键安装,无需额外配置。
Browser Automation 是免费的吗?
是的,Browser Automation 完全免费,采用 MIT-0 许可证,可自由下载、安装和使用。
Browser Automation 支持哪些平台?
Browser Automation 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 Browser Automation?
由 Charles Zhang(@tielei)开发并维护,当前版本 v0.0.5。