Claude Code Guide for Linux Mint 22.x
Complete installation, configuration, and user guide for Claude Code — Anthropic's terminal-native AI coding assistant — on Linux Mint 22.x (Cinnamon, MATE, or XFCE).
💡 Written by the same team building AI Cockpit — ▶ try the live demo, no install needed. Claude Code is one of the agents AI Cockpit orchestrates (Card 1).
About this guide: a full, community-written reference covering native installation, authentication, configuration files, slash commands, MCP servers, hooks, sandbox security/permissions, using local LLMs (Ollama) alongside Claude Code, the official web interface, and backup & migration. 35 pages in the full PDF — this page is the section overview; download the PDF below for the complete walkthrough, code samples, and the printable cheat sheet.
System requirements: Linux Mint 22.x / Ubuntu 20.04+ / Debian 10+, x86_64, 4GB RAM minimum (8GB+ comfortable, 32GB+ ideal if also running local LLMs), internet connection for Anthropic API calls. Node.js is only needed for the legacy npm install method — the native installer doesn't require it. A Claude.ai Pro/Max/Team plan or an Anthropic Console API key.
What's covered
1. What is Claude Code?
Key capabilities (reads/edits/creates files, shell command execution, full git integration, multi-agent orchestration, MCP server support, CLAUDE.md project memory, custom slash commands/hooks, background auto-updates) and the full system requirements table.
2. Pre-Installation Checklist
Verifying your shell, confirming curl (ships with Mint already), and the optional Node.js install (only needed for some MCP servers).
3. Installing Claude Code
Method A — the native installer (recommended). Method B — npm install (legacy, deprecated). Verifying the install, and migrating from a previous npm install to native.
4. Authentication
First-run browser login for Pro/Max plans, and API key authentication via the Anthropic Console for pay-as-you-go usage.
5. Configuration Files
The global settings.json starter config, and per-project CLAUDE.md — Claude Code's persistent project memory.
6. Your First Session — Step by Step
A guided walkthrough of the first real session, start to finish.
7. Slash Commands Reference
Essential built-in commands, keyboard shortcuts in interactive mode, and the different session modes.
8. Choosing a Model
Picking a model, and what happens if you switch mid-session.
9. Custom Slash Commands
Where command files live, creating one (worked example: /review), and example commands for common development tasks.
10. Hooks — Automate Actions
Hook lifecycle events, with worked examples: auto-formatting Python files after an edit, and running tests automatically after any Python edit.
11. MCP Servers — Extending Claude Code
Configuring an MCP server, and a list of useful MCP servers for a developer.
12. Security & Permissions
The permission model, recommended deny rules, and sandbox mode — what it is and how it differs from deny rules.
13. Updates & Maintenance
Auto-updates, manual updates, diagnosing problems, and a full uninstall procedure.
14. Troubleshooting
Common problems and fixes.
15. Tutorial — Practical Examples
Seven worked tutorials: understanding an unfamiliar codebase, finding and fixing a bug, adding documentation to a unit, refactoring with git safety, multi-file search, generating a test suite, and shell-scripting assistance.
16. Backup, Restore & Migration
What to back up, a backup script, restoring after a fresh install, git-based versioning (recommended), and migrating to a new machine.
17. Claude Code on the Web
The official web interface at claude.ai/code, remote sessions from the CLI, community web UIs (local browser interfaces — this is exactly what AI Cockpit is), and using the CLI and web together.
18. Using Local LLMs with Claude Code
Method A — Ollama (recommended, CLI-native). Method B — LM Studio (GUI-based model manager). Switching between local and cloud, recommended local models for your system, and using local and cloud models together.
19. Building a Personal Knowledge Library
The concept, recommended directory structure, LIBRARY.md as the master index, making the library globally available, a per-project library subset, and a strategy for adding new material over time.
20. Quick Reference Card + Printable Cheat Sheet
A condensed reference card, plus a light-theme printable cheat sheet at the end of the PDF.
Get the full guide
All 35 pages — full command syntax, code samples, and the printable cheat sheet — are in the PDF. Use the buttons at the top of the Guides page, or download directly below.
Ruggi.Site