Claude Code

Claude Code is Anthropic's agentic coding tool that lets you collaborate with Claude directly from your terminal.

Installation

See the official installation guide.

Base Invocation

claude -p "<prompt>"

Common Flags

Any with key other than provider and fallback is passed as a CLI flag. Snake-case keys are normalized to kebab-case.

Flag	Type	Description
--model	string	Model variant (e.g. sonnet, opus)
--bare	boolean	Minimal output mode
--max-turns	number	Maximum tool-calling iterations

Example

steps:
  - name: review-pr
    action: harness.run
    with:
      provider: claude
      prompt: "Review the current branch for bugs and style issues"
      model: sonnet
      bare: true

Generated invocation:

claude -p "Review the current branch for bugs and style issues" --bare --model sonnet

YOLO Mode

By default, Claude Code asks for approval before running commands. In a Dagu workflow this is not possible, so you should run in non-interactive mode:

steps:
  - name: auto-refactor
    action: harness.run
    with:
      provider: claude
      prompt: "Refactor the auth module"
      dangerously_skip_permissions: true

Generated invocation:

claude -p "Refactor the auth module" --dangerously-skip-permissions

Warning: This skips all safety checks. Only use in isolated or trusted environments.

Approval Push-back

When combined with Dagu's approval gates, Claude Code steps can be reviewed and iterated:

steps:
  - id: implement
    action: harness.run
    with:
      provider: claude
      prompt: "Implement the requested feature"
      model: opus
    approval:
      prompt: "Review the implementation"
      input: [FEEDBACK]

On push-back, the prompt is augmented with reviewer feedback and the previous stdout log path.

See Also

• Claude Code CLI Reference — Complete flag reference
• Permission Modes — How approval and YOLO mode work
• Environment Variables — CLAUDE_CODE_* env vars
• Common Workflows — Example prompts and patterns
• Settings — Global and project-level configuration