| name | browserwing |
| description | "通过 HTTP API 控制浏览器自动化：页面导航、元素交互（点击/输入/选择）、数据提取、无障碍快照、截图、JS 执行、批量操作。触发：需要通过 API 程序化控制浏览器。不适用于：一次性简单浏览器任务（用 agent-browser）、平台特定交互（用 bb-browser）。" |
BrowserWing Executor API
Browser automation through HTTP APIs — navigate, interact, extract, and analyze pages.
Configuration
- Environment Variable:
BROWSERWING_EXECUTOR_URL
- Default:
http://127.0.0.1:8080
<<<<<<< HEAD
- Auth:
X-BrowserWing-Key: <api-key> header or Authorization: Bearer <token> if required
=======
- Auth:
X-BrowserWing-Key: <api-key> header or Authorization: Bearer SANITIZED_TOKEN<token> if required
0b16965cf9e8ed0cbfb77a2dd281c8c04f090264
- Base URL:
${BROWSERWING_EXECUTOR_URL}/api/v1/executor
In shell: EXECUTOR_URL="${BROWSERWING_EXECUTOR_URL:-http://127.0.0.1:8080}"
Step-by-Step Workflow
- Get API URL: Read
$BROWSERWING_EXECUTOR_URL (fallback: http://127.0.0.1:8080)
- Discover commands:
GET /help (do this first if unsure)
- Navigate:
POST /navigate to open target page
- Analyze page:
GET /snapshot to understand structure and get RefIDs
- Interact: Use RefIDs (
@e1, @e2) or CSS selectors for click/type/select/wait
- Extract data:
POST /extract for information
- Present results: Format and show to user
Element Identification
- RefID (Recommended):
@e1, @e2 — stable across page changes, from /snapshot
- CSS Selector:
#id, .class, button[type="submit"]
- XPath:
//button[@id='login']
- Text Content:
Login, Sign Up
- ARIA Label: Automatically searched
Guidelines
- Always call
/snapshot after navigation — get page structure and RefIDs
- Prefer RefIDs over CSS selectors for reliability
- Re-snapshot after page changes to refresh RefIDs
- Use
/wait for dynamic/async content
- Use
/batch for multiple sequential operations
- If operation fails, check identifier and try different format
Route Table
