Dify - Workflow Editor

Site: dify|https://dify.ai/

Get plain text:

curl -H "Accept: text/plain" https://jugoya.ai/guides/dify/workflow-editor

Overview

Navigate and inspect Dify workflow applications, view node configurations, and understand workflow structure.

Access Studio (App List)

  1. Navigate to Dify instance URL (e.g., http://{DIFY_HOST}/apps)
  2. Page shows workspace name and app list
  3. Available navigation tabs:
    • Explore - Browse templates
    • Studio - Your apps (default view)
    • Knowledge - Knowledge bases
    • Tools - Custom tools
    • Plugins - Installed plugins

App List View

UI Elements

  • button "Y {workspace_name}'s Workspace" - Workspace selector
  • button "Create from Blank" - New empty app
  • button "Create from Template" - Use template
  • button "Import DSL file" - Import workflow definition
  • input "Search" - Filter apps
  • App cards show icon, last edit time (e.g., "45""), and "Add tags" button

App Types

Filter by type using tabs:

  • All / Workflow / Chatflow / Chatbot / Agent / Completion

Open Workflow Editor

  1. Click on app card to open
  2. URL changes to /app/{APP_ID}/workflow
  3. Editor tabs available:
    • Orchestrate - Visual workflow editor
    • API Access - API endpoints and keys
    • Logs - Execution history
    • Monitoring - Performance metrics

Workflow Canvas

Viewing Nodes

  • Nodes displayed as cards with:
    • Node type icon
    • Node name (e.g., "LLM 天気")
    • Model info (e.g., "Gemini 2.5 Flash")
    • Badge indicating mode (e.g., "CHAT")
  • Edges show connections between nodes as clickable buttons

Common Node Types

  • START - Entry point with input fields
  • LLM - Language model processing
  • Template - Text formatting
  • Output - Final result

Inspect Node Details

  1. Click on any node in canvas
  2. Side panel opens with:
    • input "Add title..." - Node name
    • input "Add description..." - Node description
    • SETTINGS tab - Configuration
    • LAST RUN tab - Previous execution

LLM Node Settings

  • "model *" - Model selector
  • "context" - Context variables
  • "SYSTEM" - System prompt text (shows character count)
  • "vision" section - For image input capability
  • "Resolution" - High/Low for vision
  • button "Add Message" - Add user/assistant messages
  • "Output Variables" - Structured output config
  • "retry on failure" - Error handling

Workflow Status

Top bar shows:

  • "Auto-Saved {time}" - Last save time
  • "Published {X} days ago" - Publication status
  • "Test Run" button with shortcut (⌥R)
  • "Publish" button - Deploy changes

Tips

  • START node shows required inputs (e.g., "image required")
  • Multiple LLM nodes can run in parallel for faster processing
  • Use "View cached variables" to debug variable values
  • Zoom percentage shown (e.g., "58%")