Praison AI Agents

1.6.33 verified Fri May 01 auth: no python

Praison AI Agents is a Python framework for building and orchestrating multi-agent systems with self-reflection capabilities. It supports various AI models (OpenAI, Anthropic, Ollama, etc.) and provides tools for complex task decomposition, agent collaboration, and iterative refinement. The current version is 1.6.33, with a release cadence of multiple updates per month.

pip install praisonaiagents

Common errors

error ModuleNotFoundError: No module named 'praisonaiagents' ↓

cause Typo or incorrect package name (e.g., 'praisonai' instead of 'praisonaiagents').

fix

Run pip install praisonaiagents (note the 'agents' suffix).

error ImportError: cannot import name 'Agent' from 'praisonaiagents' ↓

cause Old import path; Agent class was moved to submodule.

fix

Use from praisonaiagents.agent import Agent.

error TypeError: __init__() missing 1 required positional argument: 'llm' ↓

cause Passing arguments to PraisonAIAgents without keyword, or missing required config.

fix

Pass llm=os.environ.get('OPENAI_API_KEY') as a keyword argument.

error KeyError: 'OPENAI_API_KEY' ↓

cause Environment variable not set but required by default LLM provider.

fix

Set export OPENAI_API_KEY='your-key' or pass a different API key via llm.

Warnings

breaking Version 1.0.0 moved the Agent and Task classes into submodules. Direct imports from the top-level package (e.g., `from praisonaiagents import Agent`) will fail. ↓

fix Use `from praisonaiagents.agent import Agent` and `from praisonaiagents.task import Task`.

breaking The `PraisonAIAgents` class no longer accepts `llm` as a positional argument; it must be passed as a keyword argument. ↓

fix Use `PraisonAIAgents(agents=[...], tasks=[...], llm=api_key)` instead of positional.

gotcha The `llm` parameter in `PraisonAIAgents` expects an API key string or a config object, not a model name. Passing a model name directly will cause authentication errors. ↓

fix Ensure you pass `os.environ.get('OPENAI_API_KEY')` or a properly configured `LLMConfig` object.

gotcha By default, agents use 'openai/gpt-4' model. If you do not set an API key, the agent will fail silently or raise ambiguous errors. ↓

fix Set the `OPENAI_API_KEY` environment variable or pass an `llm` parameter to `PraisonAIAgents`.

deprecated The `PraisonAIAgents.initialize()` method is deprecated in favor of `start()`. ↓

fix Use `agents.start()` instead of `agents.initialize()`.

Imports

PraisonAIAgents
wrong
```
from praisonai import PraisonAIAgents
```
correct
```
from praisonaiagents import PraisonAIAgents
```
The package was previously named 'praisonai' and the import path changed.

Agent

wrong

from praisonai import Agent

correct

from praisonaiagents.agent import Agent

Agent class was moved to submodule.

Task
```
from praisonaiagents.task import Task
```
Always import from the correct submodule to avoid deprecation warnings.

Quickstart

Basic setup: create an agent, a task, and run the orchestration.

import os
from praisonaiagents import PraisonAIAgents
from praisonaiagents.agent import Agent
from praisonaiagents.task import Task

api_key = os.environ.get('OPENAI_API_KEY', '')

agent = Agent(
    name="ResearchAgent",
    llm="gpt-4",
    system_prompt="You are a helpful research assistant."
)

task = Task(
    description="Summarize the latest AI trends",
    agent=agent
)

agents = PraisonAIAgents(
    agents=[agent],
    tasks=[task],
    llm=api_key
)

result = agents.start()
print(result)