If you’re a developer looking to transform your coding workflow with genuinely intelligent AI assistance, the combination of Claude Code CLI, OpenClaw, and Visual Studio Code is one of the most powerful setups available in 2026. This isn’t your average chatbot integration — it’s a full-stack agentic development environment where AI reads your entire codebase, edits files, runs commands, manages persistent sessions, and even sends messages across platforms like Slack, Telegram, and Discord, all from inside VS Code.
In this complete guide, we walk you through every single step: from installing prerequisites and authenticating with Anthropic, to wiring up the OpenClaw MCP server, configuring your workspace with CLAUDE.md, and applying pro-level optimizations that most tutorials skip. Whether you’re brand new to AI-assisted development or upgrading an existing setup, this is the only guide you’ll need.
Looking for our other AI developer tool breakdowns? Head over to our AI Tools Directory or browse our full blog archive for more deep-dive tutorials.
Table of Contents
- What Is Claude Code CLI?
- What Is OpenClaw and Why Combine It with Claude Code?
- Prerequisites and System Requirements
- Step 1 — Install Node.js and Visual Studio Code
- Step 2 — Install Claude Code CLI
- Step 3 — Authenticate Claude Code with Anthropic
- Step 4 — Install the Claude Code VS Code Extension
- Step 5 — Set Up Your OpenClaw Gateway
- Step 6 — Connect OpenClaw to Claude Code via MCP
- Step 7 — Configure CLAUDE.md and .claudeignore
- Step 8 — VS Code Settings Optimization
- Step 9 — Test and Verify Your Full Setup
- Pro Tips and Advanced Workflows
- Common Issues and Troubleshooting
- Accessibility and Inclusive Development
- Frequently Asked Questions
What Is Claude Code CLI?
Claude Code is Anthropic’s agentic coding tool, purpose-built to operate directly within a developer’s terminal and code editor. Unlike basic AI chatbots that return code snippets for you to copy and paste, Claude Code autonomously reads your entire codebase, edits files, runs commands, installs dependencies, and executes complex multi-step development tasks — all through natural language.
The key word here is agentic. When you ask Claude Code to “add email validation to the signup form,” it doesn’t just paste you some regex. It finds the relevant files across your project, plans the changes, shows you an inline diff, and applies the edits after your approval. It can run your test suite to verify nothing broke. It can commit the changes to Git. The whole loop, from task description to verified commit, happens without you leaving VS Code.
Claude Code is available across multiple interfaces:
- Terminal CLI — The most feature-complete interface; ideal for power users and scripting
- VS Code Extension — Native graphical interface integrated directly into your IDE (the focus of this guide)
- JetBrains Plugin — For IntelliJ, PyCharm, WebStorm, GoLand, and other JetBrains IDEs
- Desktop App — Visual diff review, app preview, and connectors for GitHub, Slack, Linear, Notion
- Web Interface — Zero installation, available at claude.ai/code
All five interfaces share the same subscription, so you pay once and get access everywhere.
For the full official documentation, visit the Claude Code documentation at docs.claude.com.
What Is OpenClaw and Why Combine It with Claude Code?
OpenClaw (formerly Clawdbot) is a self-hosted AI assistant platform with an OpenAI-compatible API gateway that runs locally on your machine (default port 18789). Think of it as your personal AI operations center — a lobster-themed powerhouse (🦞) that orchestrates AI tasks across your entire digital environment.
On its own, OpenClaw is impressive. But when you connect it to Claude Code through the Model Context Protocol (MCP), the combination becomes something genuinely extraordinary. The OpenClaw MCP server exposes a rich suite of tools to Claude Code, giving your VS Code-based AI assistant capabilities that go far beyond editing source files:
- Messaging — Send messages via Telegram, WhatsApp, Discord, Slack, Signal, iMessage, and Google Chat
- Session Management — List, inspect, and communicate with other OpenClaw sessions
- Scheduling — Create and manage cron jobs for automated tasks
- Device Control — Send notifications, access camera feeds, query location, and run remote commands on paired devices
- Web Access — Search and fetch web content
- Memory — Query and store information in OpenClaw’s persistent memory system
- Text-to-Speech — Generate audio from text
- Gateway Management — Status and configuration access
What makes this genuinely exciting — as one developer put it — is the AI orchestration angle: your Claude Code session in VS Code can delegate tasks to your OpenClaw bot, which then handles everything else autonomously. It’s an AI assistant orchestrating another AI assistant. This is the frontier of agentic development, and you can have it running on your local machine today.
Our digital development services team has been running this stack for several months, and the productivity gains are substantial. Let’s get you set up.
Prerequisites and System Requirements
Before diving into the installation steps, make sure you have the following:
Required Software
- Node.js 18 or higher — Claude Code is distributed as an npm package and requires a modern Node.js runtime. Download from nodejs.org.
- Visual Studio Code 1.98.0 or later — The Claude Code extension requires a recent VS Code version. Download from code.visualstudio.com.
- Git — Recommended for version control integration and committing Claude’s changes safely.
- npm (comes with Node.js) — Used to install the Claude Code CLI globally.
Accounts and Subscriptions
- Anthropic Account with Claude Pro, Max, Team, or Enterprise subscription — Claude Code is not available on the free tier. Sign up or upgrade at anthropic.com.
- OpenClaw Account / Installation — You’ll need OpenClaw running locally. Visit openclaw.ai for installation instructions.
Windows-Specific Requirements
Claude Code CLI is primarily designed for Unix-like environments (Linux and macOS). If you’re on Windows, you have two options:
- WSL2 (Recommended) — Windows Subsystem for Linux version 2 provides full Linux compatibility and the best overall experience. See Microsoft’s WSL2 installation guide.
- PowerShell (Native Windows) — Supported but with some limitations. Use the PowerShell installation script for Claude Code.
Once you have these in place, you’re ready to begin. Let’s go step by step.
Step 1 — Install Node.js and Visual Studio Code
If you already have Node.js 18+ and VS Code 1.98.0+ installed, you can skip ahead to Step 2.
Install Node.js
Head to nodejs.org/en/download and grab the LTS (Long Term Support) version, which is always Node 18 or higher. After installation, verify it worked by opening your terminal and running:
node --version
npm --versionYou should see version numbers printed for both. If you see node: command not found, make sure your PATH includes the Node.js binary directory.
Install VS Code
Download the appropriate installer from code.visualstudio.com/download. After installation, launch VS Code and confirm you’re on version 1.98.0 or higher by going to Help → About.
Windows Users: Set Up WSL2
Open PowerShell as Administrator and run:
wsl --installRestart your machine. After reboot, open the WSL2 terminal (search “Ubuntu” or “WSL” in the Start menu) and install Node.js inside the WSL2 environment using the Node Version Manager (nvm) for the most flexible setup:
curl -o- https://raw.githubusercontent.com/nvm-sh/nvm/v0.39.7/install.sh | bash
source ~/.bashrc
nvm install --lts
nvm use --ltsStep 2 — Install Claude Code CLI
There are two ways to install Claude Code CLI. The easiest is the official install script:
Method A: Official Install Script (Recommended)
curl -fsSL https://claude.ai/install.sh | bashThe script automatically detects your operating system and handles the complete setup. For Windows PowerShell users, Anthropic provides an equivalent PowerShell command — check the official Claude Code documentation for the latest PowerShell version.
Method B: npm Global Install
npm install -g @anthropic-ai/claude-codeAfter installation via either method, verify it succeeded:
claude --versionYou should see the version number printed. If you get a “command not found” error, ensure your npm global binary path is included in your system’s PATH variable:
# Add this to your ~/.bashrc or ~/.zshrc if needed:
export PATH="$HOME/.npm-global/bin:$PATH"Then reload your shell profile:
source ~/.bashrc # or: source ~/.zshrcStep 3 — Authenticate Claude Code with Your Anthropic Account
Claude Code needs to authenticate with Anthropic before you can use it. Run the CLI to start the process:
claudeThis opens a browser window where you authorize Claude Code against your Anthropic account. For Claude Pro and Max subscribers, this OAuth flow handles everything — no manual API key management is needed.
API Key Authentication (Alternative Method)
If you prefer API key authentication (e.g., for automation or CI/CD pipelines), set your key as a secure environment variable. Never hardcode your API key or write it directly to your shell profile, as this exposes it to all child processes and logging tools.
Here are the recommended secure approaches:
# Option A: Read key from a protected file (chmod 600 recommended)
export ANTHROPIC_API_KEY="$(cat ~/.secrets/anthropic_key)"
# Option B: Use 1Password CLI (recommended for teams)
export ANTHROPIC_API_KEY="$(op read 'op://Private/Anthropic/api-key')"
# Option C: Use direnv with a .env file (never commit the .env to Git)
# In your project directory, create .env:
# ANTHROPIC_API_KEY=your-key-here
# Then run: direnv allowAfter authentication, Claude Code connects securely to Anthropic’s servers and you’re ready to start coding with AI.
Step 4 — Install the Claude Code VS Code Extension
While the CLI is powerful, the VS Code extension gives you a native, graphical interface with inline diffs, real-time change visualization, plan review mode, @-mentions for files, and conversation history — all without leaving your editor.
Install from the VS Code Marketplace
- Open VS Code
- Press Ctrl+Shift+X (Windows/Linux) or Cmd+Shift+X (Mac) to open the Extensions panel
- Search for “Claude Code”
- Find the extension published by Anthropic (there are unofficial alternatives — install only the verified Anthropic one)
- Click Install
Alternatively, visit the Claude Code listing on the Visual Studio Marketplace and click the Install button there.
Note: If the extension doesn’t appear after installation, restart VS Code or run Developer: Reload Window from the Command Palette (Ctrl+Shift+P / Cmd+Shift+P).
Opening Claude Code in VS Code
After installation, the Spark ⚡ icon appears in multiple locations across VS Code:
- Editor Toolbar — Top-right corner of the editor (only visible when a file is open)
- Activity Bar — Left sidebar, always visible; click to open the sessions list
- Status Bar — Bottom-right corner of VS Code
Click any of these to open the Claude Code panel. Start a new conversation with Ctrl+N (Windows) or Cmd+N (Mac).
Extension vs. CLI: Which Should You Use?
Both interfaces access the same Claude Code backend and use the same subscription. Here’s a quick decision guide:
- Use the VS Code Extension if you prefer point-and-click diff management, visual context, and inline file editing
- Use the Terminal CLI if you want full flag access (such as
--printfor non-interactive mode or--modelfor explicit model selection), need to pipe output to other commands, or want to integrate Claude Code into shell scripts and CI/CD pipelines
Many experienced developers use both simultaneously — the extension for focused file edits and the terminal for large refactoring tasks.
Step 5 — Set Up Your OpenClaw Gateway
Before you can connect OpenClaw to Claude Code, you need the OpenClaw Gateway running locally. The gateway is OpenClaw’s local API server (default port 18789) that acts as the bridge between Claude Code and all of OpenClaw’s capabilities.
Install OpenClaw
Visit openclaw.ai for the complete installation guide for your operating system. OpenClaw supports macOS, Linux, and Windows (via WSL2). The installation process includes:
- Downloading and installing the OpenClaw application
- Running through the initial setup wizard (model selection, messaging channels, etc.)
- Confirming the gateway is running: open your browser and navigate to
http://127.0.0.1:18789— you should see the OpenClaw gateway status page
Obtain Your Gateway Auth Token
If your OpenClaw Gateway requires authentication (recommended for security), locate your gateway token in the OpenClaw configuration settings. You’ll need this token in the next step when registering the MCP server with Claude Code.
Verify Gateway Status
curl http://127.0.0.1:18789/statusA successful response indicates your gateway is ready. If you’ve configured a non-default port, update the URL accordingly and note your custom gateway URL for the MCP configuration step.
Step 6 — Connect OpenClaw to Claude Code via MCP
This is the step that unlocks the real power of this setup. The Model Context Protocol (MCP) is the open standard that allows Claude Code to communicate with external tools and services. By registering the OpenClaw MCP server with Claude Code, you give your AI coding assistant access to the full range of OpenClaw’s capabilities from within VS Code.
Option A: Using the Helms-AI OpenClaw MCP Server (Full Tool Suite)
The Helms-AI OpenClaw MCP server exposes the complete OpenClaw gateway toolkit to Claude Code:
# Step 1: Clone the repository
git clone https://github.com/Helms-AI/openclaw-mcp-server.git
cd openclaw-mcp-server
# Step 2: Install dependencies
npm install
# Step 3: Build the server
npm run buildNow register it with Claude Code using the CLI:
# Basic registration
claude mcp add openclaw -- node /absolute/path/to/openclaw-mcp-server/dist/index.js
# With gateway auth token (recommended)
claude mcp add openclaw \
-e OPENCLAW_GATEWAY_TOKEN=your-token-here \
-- node /absolute/path/to/openclaw-mcp-server/dist/index.js
# With custom gateway URL (if not using the default port)
claude mcp add openclaw \
-e OPENCLAW_GATEWAY_TOKEN=your-token-here \
-e OPENCLAW_GATEWAY_URL=http://127.0.0.1:YOUR_PORT \
-- node /absolute/path/to/openclaw-mcp-server/dist/index.jsImportant: Always use absolute paths (e.g., /Users/yourname/... or C:\Users\... on Windows). Relative paths and ~ shortcuts can cause connection failures.
Option B: JSON Configuration (claude_desktop_config.json / VS Code settings.json)
If you prefer to configure MCP servers via JSON, add the following to your Claude configuration file:
Configuration file locations:
- macOS:
~/Library/Application Support/Claude/claude_desktop_config.json - Windows:
%APPDATA%\Claude\claude_desktop_config.json - Linux:
~/.config/Claude/claude_desktop_config.json
{
"mcpServers": {
"openclaw": {
"command": "node",
"args": ["/absolute/path/to/openclaw-mcp-server/dist/index.js"],
"env": {
"OPENCLAW_GATEWAY_TOKEN": "your-token-here",
"OPENCLAW_GATEWAY_URL": "http://127.0.0.1:18789"
}
}
}
}Restart Claude Code (or VS Code) after saving this configuration file.
Option C: OpenClaw Claude Code Skill (Sub-Agent Orchestration)
For advanced multi-agent workflows, the openclaw-claude-code-skill package enables persistent sessions, sub-agent orchestration, and fine-grained tool control:
# Install globally
npm install -g openclaw-claude-code-skill
# Set your backend API URL
export BACKEND_API_URL="http://127.0.0.1:18795"
# Start a persistent session
claude-code-skill session-start myproject -d ~/your-project \
--permission-mode plan \
--allowed-tools "Bash,Read,Edit,Write,Glob,Grep"
# Send a task to the session
claude-code-skill session-send myproject "Find all TODO comments and fix them" --streamVerifying MCP Connection
After registering the OpenClaw MCP server, open Claude Code in VS Code and type:
/pluginsThis opens the plugin management interface. You should see the OpenClaw tools listed as available. Alternatively, simply ask Claude Code: “What OpenClaw tools do you have access to?” and it will enumerate the available capabilities.
Step 7 — Configure CLAUDE.md and .claudeignore
Two configuration files dramatically improve Claude Code’s usefulness for your specific project: CLAUDE.md and .claudeignore. Skipping this step is one of the most common beginner mistakes — it’s the difference between a generic AI assistant and one that truly understands your project.
CLAUDE.md — Persistent Project Instructions
Create a CLAUDE.md file in your project’s root directory. Claude Code reads this file at the start of every session, giving it persistent context about your project:
# Project: Your Project Name
## Overview
Brief description of what this project does, its main purpose, and target users.
## Tech Stack
- Language: TypeScript / Node.js 20
- Framework: Next.js 15 (App Router)
- Database: PostgreSQL via Prisma ORM
- Styling: Tailwind CSS v4
- Testing: Vitest + Playwright
- Deployment: Vercel
## Repository Structure
- /app — Next.js App Router pages and layouts
- /components — Reusable React components
- /lib — Utility functions and shared logic
- /prisma — Database schema and migrations
- /tests — Unit and integration tests
- /public — Static assets
## Coding Conventions
- Use TypeScript strict mode; avoid 'any' types
- Components: functional, arrow functions, named exports
- API routes: follow RESTful conventions with proper HTTP status codes
- Always write a test for new utility functions
- CSS: Tailwind utility classes; avoid custom CSS unless necessary
## Git Workflow
- Branch naming: feature/description, fix/description, chore/description
- Commit format: Conventional Commits (feat:, fix:, chore:, docs:)
- Always run `npm test` before committing
## Important Notes
- Do NOT modify the /prisma/migrations folder directly
- Environment variables are in .env.local (never commit this file)
- The /scripts directory contains one-off migration scripts; do not run them automatically
You can also create CLAUDE.md files in subdirectories for folder-specific instructions. Claude Code reads them hierarchically, so global conventions in the root file combine with more specific guidance in subdirectories.
.claudeignore — Context Window Management and Security
Create a .claudeignore file (same syntax as .gitignore) to prevent Claude Code from reading sensitive or irrelevant files. This serves two critical purposes: keeping your secrets safe and keeping the context window lean for better performance:
# .claudeignore
# Secrets and credentials (CRITICAL - never expose these)
.env
.env.local
.env.production
*.pem
*.key
*.p12
*.pfx
secrets/
private/
# Build outputs and generated files
dist/
build/
.next/
out/
coverage/
# Dependencies (large and irrelevant)
node_modules/
.pnp/
.pnp.js
# Lock files (verbose and rarely useful for Claude)
package-lock.json
yarn.lock
pnpm-lock.yaml
# Database files
*.db
*.sqlite
*.sqlite3
# Logs
*.log
logs/
# OS and editor artifacts
.DS_Store
.idea/
*.swpA well-crafted .claudeignore keeps the context window focused on what matters, which directly translates to faster responses and more accurate, project-specific outputs.
Step 8 — VS Code Settings Optimization
A few targeted VS Code settings can significantly improve your Claude Code workflow. Open your settings with Ctrl+, (Windows/Linux) or Cmd+, (Mac), then click the Open Settings (JSON) icon in the top-right to edit settings.json directly:
{
// Auto-save files so Claude Code always operates on current content
"files.autoSave": "afterDelay",
"files.autoSaveDelay": 1000,
// Use VS Code's integrated terminal for Claude Code CLI access
"terminal.integrated.defaultProfile.linux": "bash",
"terminal.integrated.defaultProfile.osx": "zsh",
// Enable inline suggestions (complements Claude Code)
"editor.inlineSuggest.enabled": true,
// Claude Code Extension settings
"claudeCode.enableTelemetry": false,
// For third-party providers (e.g., Amazon Bedrock) - optional
// "claudeCode.env": {
// "CLAUDE_CODE_USE_BEDROCK": "1",
// "AWS_REGION": "us-east-2"
// },
// Disable login prompt if already authenticated
"claudeCode.disableLoginPrompt": false,
// Use terminal interface instead of graphical panel (optional preference)
// "claudeCode.useTerminal": true,
// Git: auto-fetch for better branch context
"git.autofetch": true,
"git.confirmSync": false
}Custom Keybindings for Claude Code
Open Ctrl+K Ctrl+S (or Cmd+K Cmd+S) to access keybindings, then click Open Keyboard Shortcuts (JSON):
[
{
"key": "ctrl+shift+a",
"command": "claudeCode.newSession",
"when": "editorFocus"
},
{
"key": "ctrl+shift+k",
"command": "claudeCode.insertFileReference",
"when": "editorFocus"
}
]These bindings eliminate the friction of navigating menus. The file reference shortcut (Alt+Ctrl+K on Windows/Linux or Cmd+Option+K on Mac) inserts @File#L1-99-style references so you can precisely scope Claude’s context to specific lines.
Step 9 — Test and Verify Your Full Setup
With everything installed and configured, let’s do a full end-to-end verification. Open a real project in VS Code (or create a new one with mkdir test-project && cd test-project && git init) and follow these checks:
Verify Claude Code CLI
# Check version
claude --version
# List registered MCP servers
claude mcp list
# Start Claude Code in your project
cd your-project
claudeYou should see your OpenClaw server listed. Claude Code will greet you with a prompt interface in the terminal.
Verify the VS Code Extension
- Click the Spark ⚡ icon in the VS Code Activity Bar
- The Claude Code panel opens — you should be logged in automatically
- Start a new conversation (Ctrl+N or Cmd+N)
- Type: “Describe the structure of this project”
- Claude Code should accurately describe your project’s file structure, referencing your CLAUDE.md context
Verify OpenClaw Integration
- In the Claude Code panel, type: “What OpenClaw tools are available?”
- Claude should list the OpenClaw gateway tools (messaging, sessions, scheduling, etc.)
- Test a simple OpenClaw action: “Use OpenClaw to check the gateway status”
If the OpenClaw tools don’t appear, double-check that your gateway is running (curl http://127.0.0.1:18789/status) and that the MCP server path in your configuration uses the correct absolute path. See our troubleshooting section below for common fixes.
Pro Tips and Advanced Workflows
Once your base setup is running, these advanced techniques will take your productivity to the next level.
1. The Explore-Plan-Code-Commit Workflow
The most effective Claude Code workflow follows four phases:
- Explore — Ask Claude to read and understand the relevant code before making any changes
- Plan — Enable Plan Mode (the review/approval step before Claude applies changes) to see and approve Claude’s proposed approach
- Code — Let Claude implement the changes; review each diff carefully
- Commit — Commit the verified changes to Git before starting the next task
Always commit before asking for large, multi-file changes. This gives you a clean rollback point if something goes wrong.
2. Using Plan Mode for Complex Tasks
For complex tasks, enable Plan Mode in the VS Code extension. Claude presents its plan for your review and approval before writing a single line of code. This prevents Claude from going off in the wrong direction on a multi-step task and is invaluable for architectural decisions.
3. Context Management with /clear
One of the most common beginner mistakes is running mixed-topic sessions without clearing context. Use the /clear slash command between unrelated tasks to reset the context window. This prevents earlier conversation history from polluting Claude’s focus on a new task.
4. Multi-File @-Mentions
In the VS Code extension, use @filename syntax to bring specific files into context. You can even reference specific line ranges: @src/auth/login.ts#L45-L90. This keeps the context window focused on exactly what Claude needs to complete the task.
5. OpenClaw Sub-Agent Orchestration
One of the most powerful patterns with the Claude Code + OpenClaw integration is using OpenClaw as a coordinator for parallel Claude Code tasks. You can instruct Claude Code in VS Code to delegate specific sub-tasks to OpenClaw sessions, which spin up their own Claude Code instances to handle different components simultaneously.
# In your Claude Code session:
"Use OpenClaw to start a new session that:
1. Runs our full test suite
2. Reports back the failing tests
3. We'll fix them one by one in this session"6. Automated Testing Loops
Claude Code can run your test suite, read the failure output, fix the code, and re-run the tests in a loop — all autonomously. Set this up with a simple prompt:
"Run the test suite. For each failing test, read the relevant source files,
fix the underlying issue, and re-run the specific test to verify.
Do not proceed to the next failing test until the current one passes.
Commit each fix separately with a descriptive commit message."7. CLAUDE.md Imports and Hierarchy
For large monorepos, use CLAUDE.md imports to keep instructions modular:
# In root CLAUDE.md:
@import packages/backend/CLAUDE.md
@import packages/frontend/CLAUDE.md
@import docs/architecture.md8. Non-Interactive Mode for CI/CD
Claude Code’s --print flag enables non-interactive mode, making it suitable for automation:
# In a CI/CD pipeline (GitHub Actions, etc.):
claude --print "Review this PR diff for security vulnerabilities and output a JSON report" \
< diff.txt > security-report.jsonThis integrates AI code review directly into your deployment pipeline. Check out our development automation services if you want help implementing this kind of workflow.
Common Issues and Troubleshooting
Issue: “claude: command not found”
Solution: Your PATH doesn’t include the npm global binaries directory. Run npm config get prefix to find where npm installs global packages, then add [prefix]/bin to your PATH. Reload your shell with source ~/.bashrc or source ~/.zshrc.
Issue: Claude Code extension not appearing after install
Solution: Open the Command Palette (Ctrl+Shift+P / Cmd+Shift+P) and run Developer: Reload Window. If it still doesn’t appear, verify you’re on VS Code 1.98.0 or later and that you installed the extension published by Anthropic (not an unofficial alternative).
Issue: OpenClaw MCP tools not appearing in Claude Code
Solution: Check these in order: (1) Confirm the OpenClaw Gateway is running — curl http://127.0.0.1:18789/status should return a success response. (2) Verify the MCP server path uses an absolute path, not a relative one or a ~ shortcut. (3) Run claude mcp list to confirm the server is registered. (4) Restart Claude Code after any configuration changes.
Issue: Authentication errors / “Not authenticated”
Solution: Run claude auth logout followed by claude auth login to re-trigger the OAuth flow. If using API key auth, verify your ANTHROPIC_API_KEY environment variable is set correctly in the current shell session with echo $ANTHROPIC_API_KEY.
Issue: Claude Code is slow or context window errors
Solution: Review your .claudeignore file — you may be allowing Claude to index too many files. Common culprits are node_modules, dist, build, and .git directories. Also use the /clear command between unrelated tasks to reset the context window.
Issue: Windows — WSL2 path issues
Solution: When specifying paths in MCP configuration files from WSL2, use Linux-style paths (e.g., /home/username/...) rather than Windows paths (e.g., C:\Users\...). For paths that cross the WSL2/Windows boundary, use the /mnt/c/Users/... format to reference Windows filesystem locations from WSL2.
Still stuck? Our team is happy to help — reach out via the MediaMato contact page or check the Claude Code GitHub Issues tracker for known bugs and community solutions.
Accessibility and Inclusive Development
Setting up powerful developer tools should be accessible to everyone. Here’s what you need to know about the accessibility of this stack:
VS Code has industry-leading accessibility support: full screen reader compatibility (NVDA, JAWS, VoiceOver), high contrast themes, keyboard-only navigation for all features, and configurable font sizes and zoom levels.
Claude Code terminal CLI is fully keyboard-driven and works with any terminal emulator that supports screen readers. Text-based interfaces are inherently accessible for users relying on assistive technology.
Claude Code VS Code Extension follows VS Code’s built-in accessibility standards. The graphical diff viewer includes keyboard navigation for accepting/rejecting changes without mouse interaction.
OpenClaw supports multiple communication channels including text-based messaging platforms, which can be more accessible than voice-only interfaces for users who are deaf or hard of hearing.
For teams building accessible products and applications with these tools, we’ve published a comprehensive Accessibility Compliance Guide covering WCAG 2.2 standards, ARIA implementation, automated testing integration, and how to incorporate accessibility auditing into your Claude Code workflow.
Frequently Asked Questions
What is Claude Code CLI?
Claude Code CLI is Anthropic’s agentic coding tool that operates in your terminal and IDE. It reads your entire codebase, edits files, runs terminal commands, and executes multi-step development tasks autonomously based on natural language prompts. It’s available as a standalone CLI, a VS Code extension, a JetBrains plugin, a desktop app, and a web interface at claude.ai/code.
What is OpenClaw and how does it relate to Claude Code?
OpenClaw (formerly Clawdbot) is a self-hosted AI assistant platform with an OpenAI-compatible gateway. It connects to Claude Code and other MCP-compatible clients via the Model Context Protocol (MCP), allowing Claude Code to delegate tasks, send messages across platforms (Telegram, WhatsApp, Slack, Discord), manage sessions, run scheduled jobs, control paired devices, and more — all from within VS Code.
What are the system requirements for Claude Code CLI?
Claude Code CLI requires Node.js 18 or higher, VS Code version 1.98.0 or later for the extension, and an Anthropic account with a Claude Pro, Max, Team, or Enterprise subscription. Windows users are strongly recommended to use WSL2 for the best experience, though PowerShell is also supported.
Do I need an API key to use Claude Code?
For Claude Pro and Max subscribers, Claude Code supports OAuth authentication — no manual API key management is required. For API key authentication, set the ANTHROPIC_API_KEY as a secure environment variable using your OS keychain, a secrets manager, or a direnv-managed .env file. Never hardcode API keys in shell profile files.
What is MCP and why does it matter for Claude Code + OpenClaw?
MCP (Model Context Protocol) is an open standard that allows AI models like Claude to communicate with external tools and services. For the Claude Code + OpenClaw integration, MCP acts as the bridge that gives Claude Code access to OpenClaw’s full suite of capabilities — messaging, session management, scheduling, device control, web search, memory, and more — all accessible from inside VS Code.
Can I use the VS Code extension and terminal CLI at the same time?
Yes. Both use the same subscription and can run simultaneously. Many developers use the terminal for multi-file refactoring tasks and the VS Code extension for focused, file-specific edits. Sessions are independent of each other.
What is CLAUDE.md and why should I configure it?
CLAUDE.md is a Markdown file in your project’s root directory that provides Claude Code with persistent, project-specific instructions — coding conventions, architecture overviews, preferred libraries, test frameworks, and deployment procedures. A well-crafted CLAUDE.md dramatically improves the quality of Claude Code’s output for your specific project.
How do I prevent Claude Code from reading sensitive files?
Create a .claudeignore file in your project root (same syntax as .gitignore) to exclude sensitive files and directories. Include .env files, private keys, build output directories, node_modules, lock files, and any files containing secrets or credentials. This also keeps the context window lean for better performance.
Is Claude Code + OpenClaw accessible for users with disabilities?
VS Code has strong accessibility support including screen reader compatibility and full keyboard navigation. Claude Code’s terminal CLI is fully keyboard-driven. The VS Code extension follows VS Code’s built-in accessibility standards. See our Accessibility Compliance Guide for more on building accessible workflows.
What subscription do I need to use Claude Code?
Claude Code requires a Claude Pro, Max, Team, or Enterprise subscription from Anthropic. It is not available on the free tier. The same subscription works across all Claude Code interfaces.
Conclusion: Your AI-Powered Development Workflow Starts Now
You’ve just set up one of the most capable AI-assisted development environments available in 2026. The combination of Claude Code CLI, OpenClaw, and Visual Studio Code gives you:
- An agentic AI coding assistant that understands your entire codebase
- Real-time inline diffs with full control over what gets applied
- Persistent project instructions via CLAUDE.md for consistently high-quality outputs
- Full MCP integration with OpenClaw for messaging, automation, scheduling, and device control
- The ability to orchestrate AI sub-agents for complex, parallel development tasks
The developers who will define the next generation of software products are the ones learning to work effectively with AI agents rather than just asking them questions. This setup puts you firmly in that category.
For more on how we help businesses and developers build, optimize, and launch powerful digital products using the latest AI tools, explore our full service offerings or take a look at our portfolio of recent projects. And if you’ve found this guide helpful, share it with your dev team and check out more tutorials on our blog.
Questions? Ideas for the next guide? We’d love to hear from you at our contact page.
About MediaMato
Mediamato is a digital development and AI strategy studio helping businesses build smarter, faster, and more accessibly. From AI-powered web applications to developer tooling and digital accessibility compliance, we’re at the intersection of human-centered design and cutting-edge technology. Learn more about us or get in touch today.