Basic prompt (interactive vs. one-shot)

# Start an interactive session
claude

# Start interactive with an initial prompt
claude "explain the architecture of this project"

# One-shot, print result and exit (non-interactive)
claude -p "what does src/main.py do?"

# Pipe input in
cat error.log | claude -p "summarize these errors"

With options

# Pick a model
claude --model claude-opus-4-8 -p "refactor this function"

# Continue the most recent conversation
claude --continue
claude -c -p "now add tests for that change"

# Resume a specific session by ID
claude --resume 550e8400-e29b-41d4-a716-446655440000

# Skip permission prompts (use with care)
claude --dangerously-skip-permissions -p "run the test suite"

# Control allowed tools
claude --allowedTools "Bash(git log:*)" "Read" -p "summarize recent commits"

# JSON output for scripting
claude -p "list the API endpoints" --output-format json

Using a command (subcommands)

# Set up / repair your installation
claude doctor

# Update to the latest version
claude update

# Manage MCP servers
claude mcp list
claude mcp add my-server /path/to/server

# Manage config
claude config list
claude config set -g theme dark

Combining options + command + prompt

# One-shot with a session continue + specific model
claude -c --model claude-sonnet-4-6 -p "fix the failing lint errors"

# Resume, restrict tools, output JSON (good for CI/scripts)
claude --resume <session-id> \
       --allowedTools "Read" "Bash(npm test:*)" \
       --output-format json \
       -p "verify the build passes"

Common options quick reference