Skip to main content
from praisonaiagents import Agent

agent = Agent(name="default-agent", instructions="Use the default agent framework.")
agent.start("What is the current default framework for this agent?")
The default framework changed from crewai to praisonai when running YAML files without specifying an explicit framework. 
from praisonai import PraisonAI

# No framework= argument — uses 'praisonai' by default
praison = PraisonAI(agent_file="agents.yaml")
result = praison.run()
The user runs a YAML file without framework:; agents execute on the praisonai adapter by default.

​
Quick Start

1

Before (v0.x)

Running praisonai agents.yaml would fail with error:
ValueError: Unknown praisonai.framework_adapters plugin: ''.
Available: ['ag2', 'autogen', 'autogen_v4', 'crewai', 'praisonai']
2

After (v1.0.0+)

Running praisonai agents.yaml works out of the box using the praisonai framework:
praisonai agents.yaml  # ✅ Uses praisonai framework by default

​
Agent-Centric Example

from praisonai import PraisonAI

# No framework= argument — uses 'praisonai' by default
praison = PraisonAI(agent_file="agents.yaml")
result = praison.run()
Equivalent CLI command: 
praisonai agents.yaml
Both produce the same result with no --framework flag required.

​
Framework Precedence

PraisonAI resolves the framework in this order:
PrioritySourceExample
1CLI flag--framework crewai
2YAML keyframework: crewai
3Defaultpraisonai
The set of valid framework names is discovered dynamically from the adapter registry — see Framework Adapter Plugins.

​
How to Keep CrewAI as Default

If you want to continue using CrewAI as your default framework, you have two options:

​
Option 1: YAML Configuration

framework: crewai
topic: Your task description
roles:
  # Your agents here

​
Option 2: CLI Flag

praisonai --framework crewai agents.yaml

​
User Interaction Flow

​
Best Practices

Add framework: crewai (or your preferred adapter) to shared YAML so every teammate gets the same runtime without passing CLI flags.
Keep YAML portable and pass --framework crewai only when experimenting with a different adapter.
Run praisonai --list-frameworks in CI to confirm the expected adapter is installed before kicking off agent workflows.
During upgrades, set framework: praisonai explicitly in YAML so behaviour stays predictable even if defaults change again.

CrewAI Framework

CrewAI framework integration guide

PraisonAI Agents

PraisonAI native agents framework