Principle:Infiniflow Ragflow Agent Creation
| Knowledge Sources | |
|---|---|
| Domains | RAG, Agent_Systems, Workflow_Orchestration |
| Last Updated | 2026-02-12 06:00 GMT |
Overview
A workflow initialization pattern that creates an agent canvas entity from scratch or from pre-built templates defining multi-step AI pipelines.
Description
Agent Creation establishes a UserCanvas record containing a DSL (Domain Specific Language) that defines the agent's workflow graph. RAGFlow provides 25+ pre-built templates for common patterns (customer support, data analysis, content generation, etc.). Each canvas has a title, DSL graph definition, permission level, and category (agent_canvas or dataflow).
Usage
Create an agent when building multi-step AI workflows that go beyond simple RAG chat. Templates provide starting points for common patterns.
Theoretical Basis
The agent architecture follows a graph-based DSL pattern:
- Nodes: Each node represents a component (LLM, Retrieval, Categorize, etc.)
- Edges: Connections define data flow between components
- DSL: JSON structure encoding the complete graph for persistence and execution
Related Pages
Implemented By
Page Connections
Double-click a node to navigate. Hold to expand connections.
Principle
Implementation
Heuristic
Environment