菜单
Browser automation through the repo agent-browser CLI. Explicit helper for navigation, forms, screenshots, scraping, and web-app checks. Prefer Browser Use or Playwright when available. Do NOT load for: sharing URLs, embedding links, or editing screenshot files.
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
HAR: Execute Plans.md tasks from single task to full parallel team run. Trigger: implement, execute, do everything, breezing, team run, parallel. Do NOT load for: planning, review, release, setup.
Explicit helper for authentication and payment implementation with Clerk, Supabase Auth, or Stripe. Do NOT load for: general UI work, database design, or non-auth features.
CI red? Call us. Pipeline fire brigade deploys. Use when user mentions CI failures, build errors, test failures, or pipeline issues. Do NOT load for: local builds, standard implementation work, reviews, or setup.
Explicit helper for CRUD scaffolding and API endpoint generation. Do NOT load for: UI components, form design, database schema discussion, or general implementation.
Deploy to Vercel/Netlify. One-way ticket to production arranged. Use when user mentions deployment, Vercel, Netlify, analytics, or health checks. Do NOT load for: implementation work, local development, reviews, or setup.
Generate an Acceptance Demo HTML for non-engineer vibecoders right before ship/wait/reject decision. Reads back the acceptance_criteria that were stored as personal-preference.v1 by harness-plan-brief (joined by user_request_hash), then renders a single-file HTML showing each criterion as verified or unverified along with a ship/wait/reject recommendation. Use when the user asks for an acceptance review, wants to decide whether to ship a delivered task, or says: acceptance demo, accept demo, acceptance decision, acceptance review, ship/wait/reject decision, acceptance inspection. Do NOT load for: implementation, code review, release work.
| name | agent-browser |
| description | Browser automation through the repo agent-browser CLI. Explicit helper for navigation, forms, screenshots, scraping, and web-app checks. Prefer Browser Use or Playwright when available. Do NOT load for: sharing URLs, embedding links, or editing screenshot files. |
| description-en | Browser automation through the repo agent-browser CLI. Explicit helper for navigation, forms, screenshots, scraping, and web-app checks. Prefer Browser Use or Playwright when available. Do NOT load for: sharing URLs, embedding links, or editing screenshot files. |
| allowed-tools | ["Bash","Read"] |
| user-invocable | false |
| disable-model-invocation | true |
| context | fork |
| argument-hint | [url] [--headless] |
A skill for browser automation. Uses the agent-browser CLI to perform UI debugging, verification, and automated interaction.
This skill is automatically invoked by the following phrases:
| Feature | Details |
|---|---|
| Browser Automation | See references/browser-automation.md |
| AI Snapshot Workflow | See references/ai-snapshot-workflow.md |
# Check installation
which agent-browser
# If not installed
npm install -g agent-browser
agent-browser install
| Request Type | Action |
|---|---|
| Open a URL | agent-browser open <url> |
| Click an element | Snapshot → agent-browser click @ref |
| Fill a form | Snapshot → agent-browser fill @ref "text" |
| Check state | agent-browser snapshot -i -c |
| Take screenshot | agent-browser screenshot <path> |
| Debug | agent-browser --headed open <url> |
For most operations, first take a snapshot and then interact using element references:
# 1. Open the page
agent-browser open https://example.com
# 2. Take a snapshot (AI-optimized, interactive elements only)
agent-browser snapshot -i -c
# Sample output:
# - link "Home" [ref=e1]
# - button "Login" [ref=e2]
# - input "Email" [ref=e3]
# - input "Password" [ref=e4]
# - button "Submit" [ref=e5]
# 3. Interact via element references
agent-browser click @e2 # Click Login button
agent-browser fill @e3 "user@example.com"
agent-browser fill @e4 "password123"
agent-browser click @e5 # Submit
# Check current state via snapshot
agent-browser snapshot -i -c
# Or check the URL
agent-browser get url
# Take a screenshot
agent-browser screenshot result.png
| Command | Description |
|---|---|
open <url> | Open a URL |
snapshot -i -c | AI-optimized snapshot |
click @e1 | Click an element |
fill @e1 "text" | Fill a form |
type @e1 "text" | Type text |
press Enter | Press a key |
screenshot [path] | Take a screenshot |
close | Close the browser |
| Command | Description |
|---|---|
back | Go back |
forward | Go forward |
reload | Reload |
| Command | Description |
|---|---|
get text @e1 | Get text |
get html @e1 | Get HTML |
get url | Current URL |
get title | Page title |
| Command | Description |
|---|---|
wait @e1 | Wait for element |
wait 1000 | Wait 1 second |
| Command | Description |
|---|---|
--headed | Show browser |
console | Console logs |
errors | Page errors |
highlight @e1 | Highlight element |
Manage multiple tabs/sessions in parallel:
# Specify a session
agent-browser --session admin open https://admin.example.com
agent-browser --session user open https://example.com
# List sessions
agent-browser session list
# Operate in a specific session
agent-browser --session admin snapshot -i -c
| Tool | Recommendation | Use Case |
|---|---|---|
| agent-browser | ★★★ | First choice. Powerful AI-optimized snapshots |
| chrome-devtools MCP | ★★☆ | When Chrome is already open |
| playwright MCP | ★★☆ | Complex E2E testing |
Principle: Try agent-browser first; use MCP tools only if it doesn't work.
--headed to show the browserclosed