---
name: "Claude Code Advanced Usage"
description: "Advanced Claude Code (claude.ai/code) usage — manage sessions, use plan mode, connect repos, schedule tasks, handle multiple projects. Complements the basic background tasks skill."
version: "1.0.0"
author: "skynet"
category: "dev"
agents: ["claude-code"]
tags: ["claude", "code", "advanced", "browser-automation", "sessions"]
tools_required: ["chrome-mcp"]
---

# Claude Code Advanced Usage

# Claude Code Advanced Usage

Browser automation skill for claude.ai/code — advanced session and task management.

## Navigation

Navigate to `claude.ai/code` to access the Claude Code interface.

## Core Interface Elements

### Sidebar
- **New session** button
- **Scheduled** link — access scheduled tasks
- **Project filter** — "All projects" dropdown to filter by project
- **Session history** — past sessions with status indicators (PR icons, change counts like +9 -0)

### Main Area
- **Prompt input** — contentEditable div for task descriptions
- **Toggle menu button**
- **Toggle plan mode button** — switch between plan and execute modes
- **Model selector** — choose between Opus 4.6 (thorough), Sonnet (fast), etc.
- **Submit button**
- **Repo selector** — "Select a repository" to search and connect GitHub repos
- **Delete button** — remove current session

## Key Features

### 1. Session Management
- View, filter, and organize past sessions
- Sessions show PR icons and change counts for completed work
- Filter by project using the dropdown

### 2. Plan Mode
- Toggle plan mode to have Claude plan before executing
- Useful for complex tasks — review the plan before committing to execution
- Toggle button is in the main area

### 3. Scheduled Tasks (claude.ai/code/scheduled)
- Set up recurring automated tasks
- Tasks run on a schedule without manual intervention
- Great for maintenance, monitoring, and regular updates

### 4. Repository Connection
- Search and select GitHub repos via the repo selector
- Claude Code works directly with your codebase
- Branch selection available for targeting specific branches

### 5. Model Selection
- **Opus 4.6** — thorough, best for complex tasks
- **Sonnet** — fast, good for simpler tasks
- Select based on task complexity and urgency

### 6. Multiple Sessions
- Run parallel sessions for different tasks
- Each session is independent with its own context

### 7. Session URLs
- Draft sessions: `claude.ai/code/draft_xxx`
- Active sessions: `claude.ai/code/session_xxx`

## IMPORTANT: Prompt Input

The prompt input is a **contentEditable div** that is NOT in the accessibility tree. You must use JavaScript to set text and submit:
```javascript
// Use JavaScript to interact with the prompt input
// Standard form_input will NOT work
// Click the element, then use keyboard input or JS to set value
```

## No Project System Warning

When automating Claude Code via browser, agents must include full context in every prompt:
- Who the user is and their account details
- What they are working on and why
- The complete goal and any relevant history
- All necessary URLs, credentials references, and prior results

Never assume the agent remembers anything from previous interactions.

## Running on Mac Minis

These browser automation tasks run on Mac Minis via mac-control on port 8200:
- **Vault**: 192.168.86.27:8200
- **Bots**: 192.168.86.50:8200
- **Jarvis**: 192.168.86.51:8200

Use the mac-control API to send browser automation commands. Example:
```
POST http://192.168.86.27:8200/execute
```