Steward

Dagu Steward is an LLM-powered assistant integrated into the Web UI. It can read, create, and modify your workflows through a chat interface with tool-calling capabilities.

Quick Setup

1. Enable Steward — Toggle it on in the Web UI at /agent-settings, or set the environment variable DAGU_AGENT_ENABLED=true.

2. Add a model — Click Add Model in the settings page and configure an LLM provider. Supported providers: anthropic, openai, openai-codex, gemini, openrouter, zai, local (Ollama, vLLM, etc.).

   For openai-codex, connect the ChatGPT Plus/Pro subscription in the model form before saving the model.

3. Set a default model — Click the star icon next to a model to make it the default.

Once configured, click the Steward button at the bottom-left corner of any page to start chatting.

If you are using Ollama or another local model server, read Local AI before setting the Base URL. Dagu expects an OpenAI-compatible base such as http://localhost:11434/v1, not a native Ollama endpoint like /api/generate.

For the full built-in steward configuration surface, start with Steward Settings. The settings docs are split into focused pages for models, tool policy, personality, and web search.

Available Tools

Steward can use these built-in tools. Some are only available when the corresponding feature is configured:

Tool	Description
bash	Execute shell commands (120s default timeout, 600s max)
read	Read file contents with line numbers
patch	Create, edit, or delete files
think	Record reasoning without side effects
navigate	Open pages in the Dagu UI
ask_user	Prompt the user with options or free-text input
delegate	Spawn sub-agents for parallel tasks
remote_agent	Delegate tasks to agents on remote nodes (when remote nodes are configured)
list_contexts	List available remote nodes for remote_agent (when remote nodes are configured)

Provider-native web search is configured in model and steward settings rather than exposed as a separate callable tool.

Tools can be individually enabled or disabled in Tool Permissions & Bash Policy.

Agent in Workflows

You can use AI capabilities directly in your DAG steps in two ways.

Agent Step (type: agent)

A multi-turn tool-calling loop — the agent reads files, runs commands, edits code, and iterates until the task is complete:

steps:
  - id: fix_config
    type: agent
    messages:
      - role: user
        content: |
          Fix the invalid database_url in /etc/app/config.yaml
    output: RESULT

Chat Step (type: chat)

A single-shot LLM call — send a prompt and get a response, no tool use:

steps:
  - id: summarize
    type: chat
    llm:
      provider: openai
      model: gpt-4o
    messages:
      - role: user
        content: "Summarize today's error logs."
    output: SUMMARY

AI Coding Tool Integration

Install the Dagu skill for external AI coding tools (Claude Code, Codex, Gemini CLI, etc.) so they can write correct Dagu DAG files.

Use Dagu's built-in installer:

dagu ai install --skills-dir ~/.agents/skills

Or use the shared skills CLI:

npx skills add https://github.com/dagucloud/dagu --skill dagu

See ai in CLI Commands for more details.

See Also

• Steward Documentation — Complete guide to Steward and its configuration
• Steward Settings — Start here for the built-in Web UI steward settings
• Models & Providers — Add models and set the default model
• Tool Permissions & Bash Policy — Control tools and bash rules
• Personality & Web Search — Configure profiles and provider-native search
• Agent Step — Using the agent as a workflow step
• Steward Tools Reference — Detailed tool parameter documentation
• Workflow Operator — Use the built-in steward from Slack or Telegram
• Basic Chat — Single-shot LLM calls in workflows