Claude Browser — Chrome Extension Automation
Personal browser automation using Claude's Chrome extension MCP tools. Operates in the user's actual browser session with full access to logged-in state, cookies, and extensions.
When to Apply
Positive Triggers
- Personal browser automation (logged-in sessions)
- Form filling in authenticated web apps
- Reading content from pages requiring login
- Interactive demonstrations and walkthroughs
- GIF recording of browser workflows
- Tab management and multi-page navigation
Negative Triggers (Do NOT Apply)
- CI/CD pipeline testing (use
playwright-browser skill) - Parallel test execution (Chrome extension is single-session)
- Headless browser operations
- Performance/load testing
Core Principles
1. Single-session — one browser, one user, one context at a time 2. User's actual browser — leverages existing login state and cookies 3. Personal context — for tasks that require the user's authenticated session 4. Always capture evidence — use gif_creator or upload_image for visual records 5. Tab-aware — always call tabs_context_mcp at session start
Navigation + Reading
| Tool | Purpose |
|---|
navigate | Navigate to URL in current tab |
read_page | Read full page content (structured) |
get_page_text | Get plain text content |
find | Search for text/elements on page |
Interaction
| Tool | Purpose |
|---|
form_input | Fill form fields |
computer | Mouse/keyboard simulation |
javascript_tool | Execute JavaScript in page |
Tab Management
| Tool | Purpose |
|---|
tabs_context_mcp | Get current tab info (CALL FIRST) |
tabs_create_mcp | Open new tab |
resize_window | Resize browser window |
switch_browser | Switch between browser profiles |
Visual Capture
| Tool | Purpose |
|---|
gif_creator | Record multi-step interaction as GIF |
upload_image | Capture and upload screenshot |
Console + Network
| Tool | Purpose |
|---|
read_console_messages | Read browser console (use pattern param to filter) |
read_network_requests | Inspect network activity |
Shortcuts
| Tool | Purpose |
|---|
shortcuts_list | List available keyboard shortcuts |
shortcuts_execute | Execute a keyboard shortcut |
Plan Management
| Tool | Purpose |
|---|
update_plan | Update task plan visible in Chrome extension |
Session Startup Protocol
1. Call tabs_context_mcp to discover current tabs 2. Only reuse a tab if the user explicitly asks to work with it 3. Otherwise create a new tab with tabs_create_mcp 4. If tab errors occur, call tabs_context_mcp to refresh
Limitations
- Single session: Cannot run parallel browser instances
- Requires `--chrome` flag: Claude Code must be launched with Chrome integration
- Not CI-suitable: Requires a visible browser with user interaction
- Dialog risk: Avoid triggering
alert(), confirm(), prompt() — they block the extension - Tab ID scope: Never reuse tab IDs from a previous session
GIF Recording Best Practices
- Capture extra frames before and after actions for smooth playback
- Name files descriptively:
login_flow.gif, form_submission.gif - Record the full workflow, not just the end state
Anti-Patterns
- Using Claude Chrome for CI/CD testing — use Playwright instead
- Triggering JavaScript alerts — they block all further commands
- Reusing stale tab IDs from prior sessions
- Running without calling
tabs_context_mcp first - Attempting parallel operations in a single-session tool
Setup Matt Pocock SkillsScaffold the per-repo configuration that Matt Pocock’s engineering agent skills rely on so they understand the issue tracker, triage labels, and domain documentation la495k196k
Lark AppsConnect Claude, Cursor or custom agents directly to Lark (Feishu) for messaging, document automation, approval workflows and enterprise data access.402k
CavemanSlash token usage by roughly 75% while keeping every technical detail intact when working with Claude Code, Cursor or similar agents.391k94.7k
Lark Skill MakerQuickly turn any Lark/Feishu OpenAPI call or multi-step workflow into a reusable agent skill with its own SKILL.md.386k16k
Running Claude Code Via Litellm CopilotRun Claude Code at a fraction of the cost by routing requests through LiteLLM to the GitHub Copilot Chat API.270k72
Codex PetGenerate a complete Codex Pet spritesheet and metadata from one reference image without needing an OpenAI key or Codex Pro.246k10