praisonai-frameworks 0.1.10

Optional agent framework adapters for PraisonAI (CrewAI, AutoGen, extensions)

PraisonAI Frameworks

Optional agent framework adapters for PraisonAI.

Implements the praisonai.framework_adapters entry-point group so YAML framework: values (crewai, autogen, …) resolve without bloating the core SDK or wrapper wheel.

Install

# Native PraisonAI only (no third-party frameworks)
pip install praisonaiagents praisonai

# CrewAI backend
pip install praisonai-frameworks[crewai]

# AutoGen v0.2 backend
pip install praisonai-frameworks[autogen]

# LangGraph backend
pip install praisonai-frameworks[langgraph]

# OpenAI Agents SDK backend
pip install praisonai-frameworks[openai-agents]

# Agno backend
pip install praisonai-frameworks[agno]

# Google ADK backend
pip install praisonai-frameworks[google-adk]

# Pydantic AI backend
pip install praisonai-frameworks[pydantic-ai]

# AutoGen v0.4 backend (autogen-agentchat / autogen-ext)
pip install praisonai-frameworks[autogen-v4]

Selecting an AutoGen version

Use framework: autogen and set AUTOGEN_VERSION to choose a backend:

AUTOGEN_VERSION	Backend
v0.2	AutoGen v0.2 (autogen)
v0.4	AutoGen v0.4 (autogen-agentchat)
ag2	AG2 fork
auto (default)	First available: v0.2 → v0.4 → ag2

Usage

# agents.yaml
framework: crewai
topic: Research AI trends
roles:
  researcher:
    role: Research Analyst
    goal: Find accurate information
    backstory: Expert researcher
    tasks:
      research:
        agent: researcher
        description: Research {topic}
        expected_output: A concise summary

praisonai run agents.yaml

Architecture

• Depends on praisonaiagents only (protocol + base helpers in praisonaiagents.frameworks)
• Registers adapters via setuptools entry points — no wrapper import required
• Lazy-imports CrewAI / AutoGen inside run() only

Adding a framework

See examples/third_party_adapter/ and docs/adding-a-framework.md.