Skip to content

Claude Code

  • https://code.claude.com/docs/en/overview
  • Claude Code is an agentic assistant that runs in your terminal.
  • Claude Code serves as the agentic harness around Claude LLM: it provides the tools, context management, and execution environment that turn a language model into a capable coding agent
  • Subscriptions: Pro, Max, Teams, or Enterprise
brew install --cask claude-code

Configuration

# Auth config can only be set via environment variables
export ANTHROPIC_BASE_URL="https://myhost.com/anthropic"
export ANTHROPIC_AUTH_TOKEN="..."
export ANTHROPIC_CUSTOM_HEADERS="x-llm-application-name:claude_code"

Agentic Loop

  • The Agentic Loop is composed of "Gather context", "Take action" & "Verify results"

  • You start the loop with your prompt and you can interrupt, steer or add context in the middle of the loop

  • Permission Modes

  • Default: claude asks before file edits and shell commands
  • Accept edits: claude edits files without asking, still asks for commands (you can still allow specific commands at the config file so claude won't ask each time)
  • Plan mode: claude uses read-only tools only
  • "Shift+Tab" to switch modes

Sessions

  • Also know as conversations
  • Sessions are tied to your current directory
  • Sessions are ephemeral, it has no persistent memory between sessions

  • If you want Claude to know something across sessions, put it in your CLAUDE.md.

  • Memory (CLAUDE.md)

  • ~/.claude/CLAUDE.md (user)

  • ./CLAUDE.md (project)

  • Context Window

  • Holds your conversation history, file contents, command outputs, CLAUDE.md, loaded skills, and system instructions
  • As you work, context fills up and claude compacts it automatically (detailed instructions from early conversations may be lost)
  • Put persistent rules in CLAUDE.md
  • Use /context to see what's using space
  • Claude Sonnet for instance has a limit of 200k tokens

Tools

  • With tools Claude Code can act
  • read your code
  • edit files
  • run commands
  • search the web
  • interact with external services

  • Each tool use returns information that feeds back into the agentic loop, informing Claude's next decision.

  • Categories of built-in tools (base capabilities)

  • File operations: Read files, edit code, create new files, rename and reorganize
  • Search: Find files by pattern, search content with regex, explore codebases
  • Execution: Run shell commands, start servers, run tests, use git
  • Web: Search the web, fetch documentation, look up error messages

  • Code intelligence: See type errors and warnings after edits, jump to definitions, find references (requires code intelligence plugins)

  • You can extend the base capabilities with

  • Skills: what claude knows
  • MCP: connect to external services
  • Hooks: automate workflows
  • Subagents: offload tasks

Commands

  • /login: log-in to add new credentials (or use the ANTHROPIC_AUTH_TOKEN env)
  • /theme
  • /clear: clear conversation history
  • /model: switch between models (claude models only)
  • /doctor: diagnoses common issues with your installation
  • /config

  • /help

  • /init: create CLAUDE.md (similar to AGENTS.md)

  • /agents: helps you configure custom subagents
  • /memory: open the memory file
  • /resume: jump back to previous conversation. Same as "claude -r"
  • /compact: define rules for the context window compactation, e.g., "/compact focus on the API changes"
  • /context: check what is using space on the context window
  • /permissions

CLI

claude # interactive mode in cwd

# Start a conversation with a prompt (task)
claude "fix the build error" # Interactive
claude -p "explain this $CONTEXT" # (print) Non-interactive (print and exit - useful for pipes)
tail -f app.log | claude -p "Slack me if you see any anomalies appear in this log stream"

# Pick a conversation
claude -r # (resume) pick session from a list
claude -c # (continue) last session in cwd
claude -c --fork-session # fork the last session and leave the last intact

mcp

  • Scopes
  • User and local scope: ~/.claude.json (in the mcpServers field or under project paths)
  • Project scope: .mcp.json in your project root (checked into source control)

  • Managed: managed-mcp.json in system directories (see Managed MCP configuration)

  • Many cloud-based MCP servers require authentication. Claude Code supports OAuth 2.0 for secure connections, it should be done with the command /mcp

claude mcp list
claude mcp get "github"
claude mcp remove "github" -s local

# HTTP Server
claude mcp add --transport http github https://api.githubcopilot.com/mcp/ # for the current project (cwd) only (local scope) - default behavior
claude mcp add --transport http --scope user github https://api.githubcopilot.com/mcp/ # for all the projects (user scope)

# SSE (Server-Sent Events) Server
claude mcp add asana --transport sse https://mcp.asana.com/sse --header "X-API-Key: your-key-here"

# Local stdio Server
claude mcp add --transport stdio --env AIRTABLE_API_KEY=YOUR_KEY airtable -- npx -y airtable-mcp-server
claude mcp add --transport stdio --env KEY=value myserver -- python server.py --port 8080
claude mcp add nu-mcp $HOME/dev/nu/nu-mcp/run.sh

# From JSON (exactly how it will be added to the config file)
claude mcp add-json weather-api '{"type":"http","url":"https://api.weather.com/mcp","headers":{"Authorization":"Bearer token"}}'

// ~/.claude.json
{ // ...
  "mcpServers": {
    "nu-mcp": {
      "type": "stdio",
      "command": "/Users/myself/mymcp/run.sh",
      "args": [],
      "env": {}
    },
    "atlassian": {
      "type": "sse",
      "url": "https://mcp.atlassian.com/v1/sse"
    },
    "github": {
      "type": "http",
      "url": "https://api.githubcopilot.com/mcp/"
    }
  }
}