Claude Code Setup
& Custom Agent Deployment

Claude Code is Anthropic's official agentic coding tool that runs in your terminal. It can read, write, and edit code across your entire codebase, run shell commands, search the web, manage Git workflows, and execute complex multi-step development tasks autonomously. Unlike IDE-based AI tools like Cursor or GitHub Copilot, Claude Code operates at the system level — it can manage files, run builds, deploy applications, and interact with external services through MCP servers. It is powered by Claude Sonnet or Opus and understands your project context through CLAUDE.md configuration files. For teams that want both terminal and IDE-based AI, see our Cursor + Claude Code power setup.

Claude Code requires an Anthropic API key. The Pro plan costs $20/month and gives you access to Claude Code with usage limits. The Max plan starts at $100/month for 5x Pro usage, or $200/month for 20x Pro usage — ideal for heavy development. API-based usage (BYOK) is billed per token with no monthly cap. Our setup fee is a one-time cost separate from Anthropic's subscription. For context, most production users spend $100-200/month on API usage, which replaces thousands of dollars in developer time. We help you choose the right plan during the discovery call based on your expected usage patterns and budget.

Claude Code requires Node.js 20.9 or higher (we install 22+), a modern terminal (macOS Terminal, iTerm2, Windows Terminal with WSL2, or any Linux terminal), and an Anthropic API key or Max/Pro subscription. For VPS hosting, we recommend at least 2 vCPUs, 4GB RAM, and 40GB SSD. Claude Code also runs on Mac, Linux, and Windows via WSL2.

Both work well. The npm installer (npm install -g @anthropic-ai/claude-code) requires Node.js 20.9+ and is the most common method. The native installer downloads a standalone binary without requiring Node.js. We recommend npm for developers who already have Node.js, and the native installer for non-technical users or minimal environments. We configure whichever works best for your setup.

Pro ($20/month) gives you Claude Code access with standard usage limits — good for occasional use. Max ($100-200/month) provides 5-20x the usage of Pro, plus access to Claude Opus — ideal for daily development. API (BYOK) lets you use your own API key with pay-per-token pricing and no monthly limits — best for heavy automation and CI/CD pipelines. The key distinction is predictability versus flexibility: Pro and Max have fixed monthly costs with usage caps, while API scales with actual usage but has no ceiling. For teams running automated pipelines that process code 24/7, API is usually more cost-effective despite variable pricing. We help you choose the right plan during the discovery call.

Standard setup gives you a clean, professional Claude Code installation with security hardening, tmux persistence, and basic MCP integrations. The Custom Agentik OS setup includes everything in Standard plus 243 specialized AI agents, 190+ custom skills, AISB multi-agent orchestration, Telegram bot integration, automated content pipelines, security scanning tools, and enterprise-grade CI/CD automation. Think of Standard as a professional workstation and Custom as a fully equipped AI operations center.

Yes. VPS is our recommended deployment for always-on access. We deploy on Hetzner (from $4/month), DigitalOcean ($6/month), AWS Lightsail ($5/month), Oracle Cloud (free tier with 4 ARM CPUs and 24GB RAM), or any Linux VPS you provide. The setup includes SSH hardening, firewall configuration, tmux persistence so Claude Code survives disconnections, and daemon scripts for auto-restart. A VPS deployment means your Claude Code instance runs 24/7 and is accessible from any device via SSH or Telegram, unlike local installations that stop when your computer sleeps.

Telegram integration is part of our Custom Agentik OS setup. We configure a Telegram bot via @BotFather, connect it to your Claude Code instance through our custom bridge, and set up command routing so you can send tasks, receive results, and monitor builds from your phone. This is not a standard Claude Code feature — it is part of our custom agent ecosystem. For more on messaging platform integrations, see our OpenClaw setup service which supports 10+ messaging platforms.

Yes. We deploy standardized Claude Code configurations across your entire team. Each developer gets the same agents, skills, CLAUDE.md rules, and Git workflows. We also set up shared MCP servers, consistent project templates, and team-wide automation pipelines. This ensures every team member has the same AI-powered development experience. Team deployments typically take 1-2 hours per developer for Standard setup, or we can create a master configuration that replicates across machines in minutes using our deployment scripts.

AISB is our multi-agent orchestration system built on top of Claude Code. It consists of 12 specialized orchestration agents that automatically route tasks to the right specialist from our pool of 243 agents. It handles intent classification, task decomposition, parallel execution, code auditing, and knowledge management. Think of it as the nervous system that makes all 243 agents work together intelligently.

CLAUDE.md is Claude Code's configuration file — it tells Claude about your project, coding standards, and preferred workflows. Custom agents are defined as specialized configurations (in ~/.claude/agents/) that give Claude Code domain-specific knowledge, tools, and instructions. For example, our react-specialist agent knows React best practices, our seo-auditor agent runs comprehensive SEO checks, and our penetration-tester agent runs security scans.

Claude Code runs on Windows through WSL2 (Windows Subsystem for Linux). We install Ubuntu on WSL2, configure Node.js, install Claude Code, and set up your development environment. It works with Windows Terminal, VS Code's integrated terminal, and any WSL2-compatible editor. Performance is identical to native Linux.

Yes. Claude Code runs natively on macOS via Terminal, iTerm2, Warp, or any terminal emulator. We install Node.js via Homebrew or nvm, configure Claude Code, set up tmux for session persistence, and integrate with your existing development tools. Works on both Intel and Apple Silicon Macs.

MCP (Model Context Protocol) servers extend Claude Code's capabilities. We configure servers based on your workflow: GitHub (repository management), filesystem (file operations), Brave Search (web research), PostgreSQL or MongoDB (database access), Stripe (payment operations), Gmail (email), Google Calendar, Slack, and custom API integrations. Each MCP server is tested end-to-end before handoff. For advanced MCP configuration or custom server development, see our dedicated MCP setup service.

Standard Claude Code setup takes 1-2 hours. Custom Agentik OS setup takes 2-4 hours depending on the number of integrations, custom agents, and automation pipelines. Both include full testing and verification — nothing is handed off until everything works. Same-day turnaround guaranteed.

Yes. Claude Code runs locally on your machine or VPS — your code never leaves your environment unless you explicitly send it to Anthropic's API for processing. With VPS deployments, we configure SSH key-only authentication, UFW firewall, fail2ban, non-root users, and encrypted connections. Your API keys are stored securely and never shared.

Our skills library contains 190+ pre-built automation recipes that Claude Code can execute. Skills cover SEO auditing, blog generation, social media posting, security scanning, code review, deployment automation, database migrations, performance testing, accessibility auditing, and dozens more. Each skill is a tested, production-proven workflow that runs with a single command.

Claude Code with our 243-agent ecosystem handles a significant portion of development, testing, deployment, content, and operations work. It does not replace human judgment for product decisions, complex architecture, or business strategy — but it can reduce a 10-person team's workload to 1-2 people. Our clients typically ship 5-10x faster with Claude Code than with traditional teams.

You can install Claude Code yourself in about 5 minutes with npm. But getting it production-ready — SSH hardening, tmux persistence, CLAUDE.md configuration, MCP servers, custom agents, skills library, Git workflows, Telegram integration, and CI/CD pipelines — takes days of trial and error. We have done this dozens of times and our 243-agent ecosystem is battle-tested across real production environments. You skip the learning curve entirely. Pair it with our Cursor setup for the ultimate AI development environment, or explore custom MCP server development for advanced tool integrations.