AI agent
Dagu AI agent is the assistant built into the Web UI. It helps create workflows, explain failures, inspect files, and automate repetitive operator tasks from inside Dagu.
Where You Use It
The same assistant foundation appears in three places:
- the Web UI assistant
- workflow steps with
action: agent.run - Workflow Operator integrations such as Slack, Telegram, Discord, and LINE
Quick Start
- Open AI agent Settings at
/agent-settings - Enable AI agent
- Add at least one model and set a default
- Open the AI agent button in the lower-left corner of the UI
For subscription-based Codex setup, see OpenAI Subscription.
What The Web UI AI agent Can Help With
Common uses:
- create or edit DAG YAML
- explain failed runs and logs
- suggest retries, parameter changes, or debugging steps
- navigate to the right page in Dagu
- work with local files when the tool policy allows it
Working In Chat
Sessions
The chat sidebar lets you reopen earlier conversations or start a new one with the + button.
Current DAG Context
When you open AI agent from a DAG or run page, Dagu automatically gives the assistant the relevant workflow context. You can also attach additional DAGs manually from the chat input.
Model And Profile
In the chat composer you can override:
- the model used for the next request
- the profile (
soul) used for personality and instructions
Safe Mode
The shield toggle controls whether commands that require confirmation should stop and ask before running.
Settings Map
Use these pages for detailed setup:
Tool Access
Depending on your settings, AI agent can work with tools such as:
- file read and edit tools
- bash command execution
- web search and page extraction
- page navigation
- user prompts
- delegation to other agents or remote nodes
You control this from Tool Permissions & Bash Policy.
Use The Agent In Workflows
When you want the assistant to run as part of a DAG, use action: agent.run.
Typical examples:
- generate a summary from logs
- review code or configuration
- prepare a release note or runbook update
- classify or route work before later steps