CrewAI
How to Visualize Your CrewAI Workflow
CrewAI is a popular multi-agent orchestration framework that lets you build teams of AI agents with defined roles, tools, and handoff routes. As your crew grows more complex, understanding who hands off to whom — and why — becomes increasingly difficult from code alone. A CrewAI visualizer like AgentLens turns your agent configuration into an interactive graph, making it easy to audit routes, spot unreachable agents, and share your architecture with teammates in seconds.
What You'll Need
- AgentLens — free, no login required, runs entirely in your browser
- Your CrewAI YAML config — or use the example below to try it out immediately
Step 1 — Write Your CrewAI YAML
AgentLens uses a straightforward YAML schema. A workflow file has two top-level keys: agents and routes.
Each agent requires an id (no spaces), a name, and optionally a model (common values: opus, sonnet, haiku) and a list of tools. Each route requires an id, source, target, and a label describing the handoff. An optional condition field annotates branching logic.
Here is a complete example modelling a four-agent CrewAI research pipeline:
version: "1.0.0"
name: "CrewAI Research Pipeline"
agents:
- id: crew_manager
name: "Crew Manager"
model: opus
tools: [TaskAssignment, Coordination]
description: "Orchestrates the research crew and assigns tasks"
- id: web_researcher
name: "Web Researcher"
model: sonnet
tools: [WebSearch, BrowserTool]
description: "Searches the web for relevant information"
- id: data_analyst
name: "Data Analyst"
model: sonnet
tools: [DataAnalysis, ChartGenerator]
description: "Analyzes collected data and generates insights"
- id: writer
name: "Content Writer"
model: haiku
tools: [ContentGeneration]
description: "Writes the final report from analyzed data"
routes:
- id: r1
source: crew_manager
target: web_researcher
label: "Assign research task"
- id: r2
source: web_researcher
target: data_analyst
label: "Pass raw data"
- id: r3
source: data_analyst
target: writer
label: "Pass insights"
- id: r4
source: writer
target: crew_manager
label: "Submit report"Step 2 — Paste into AgentLens
Open AgentLens by clicking the button below. You will see a split-pane editor: paste your YAML into the left panel. The graph on the right updates automatically within 150 ms — no save button, no page reload. The layout engine arranges nodes and edges so nothing overlaps.
Step 3 — Read the Graph
Each box in the canvas represents one agent from your agents list. The arrows between boxes correspond to your routes, with the label value shown on each edge. Node color indicates the assigned model — making it easy to spot which agents use your most capable (and most expensive) models. Click any node to open the inspector panel and see the full agent configuration.