AgentArea Agents SDK
Independent SDK for Agent Networks, Task Orchestration, and Distributed Runners
AgentArea Agents SDK enables building scalable agent networks with task-driven workflows, featuring zero dependencies, multi-provider LLM support, extensible tools, and integration with independent runners like Temporal, Restate, or Dapr for distributed execution.
✨ Key Features
- 🌐 Agent Networks: Multi-agent collaboration with event-driven communication and orchestration
- 📋 Task Management: Comprehensive task creation, assignment, progress tracking, and evaluation
- 🏃 Independent Runners: Extensible base for distributed execution using Temporal, Restate, Dapr, or custom runners
- 🤖 Multi-Provider LLM Support: Unified interface for OpenAI, Claude, Ollama, and 100+ models via LiteLLM
- 🛠️ Extensible Tool System: Built-in tools for calculations, MCP integration, human-in-loop, and task operations
- ⚡ ReAct Framework: Structured reasoning and acting with streaming support for networked agents
- 🔒 Type Safety: Comprehensive Pydantic models and type hints throughout
- 🚀 Async/Await Ready: Full asynchronous support for efficient, distributed execution
- 📊 Comprehensive Testing: Unit, integration, and distributed workflow tests with 50%+ coverage
- 📚 Developer Friendly: Clean architecture, detailed docs, and easy integration for scalable systems
💬 Contact
Welcome to join our community on
| Discord | |
|---|---|
| opensource@agentarea.ai |
📋 Table of Contents
- 🚀 Quick Start
- 🌐 Agent Networks
- 📋 Task Management
- 🏃 Distributed Runners
- 📚 Examples
- 🏗️ Architecture
- 🔌 Components
- 📖 Supported LLM Providers
- 🧪 Testing
- 💻 Development
- 🤝 Contributing
- 📄 License
- 👥 Contributors
🚀 Quick Start
Prerequisites
- Python 3.11 or higher
- pip package manager
Installation
From PyPI (when published):
pip install agentarea-agents-sdk
From source (development):
# Clone the repository
git clone https://github.com/agentarea/agentarea-agents-sdk.git
cd agentarea-agents-sdk
# Install in editable mode with dev dependencies
pip install -e .[dev]
Basic Agent Network Example
Create a simple agent network with task assignment.
import asyncio
from agentarea.agent_kit.agents.network import AgentNetwork
from agentarea.agent_kit.agents import create_agent
from agentarea.agent_kit.tasks.tasks import create_task
async def main():
# Create agents in a network
network = AgentNetwork()
math_agent = create_agent(
name="Math Agent",
instruction="You are a math specialist.",
model="ollama_chat/qwen2.5"
)
coordinator = create_agent(
name="Coordinator",
instruction="Assign tasks to specialists.",
model="ollama_chat/qwen2.5"
)
network.add_agent(math_agent)
network.add_agent(coordinator)
# Create and assign a task
task = create_task(
description="Calculate 25 * 4 + 15",
assignee="Math Agent"
)
network.assign_task(task)
# Run the network
async for event in network.run():
print(f"Network event: {event}")
asyncio.run(main())
Task Orchestration Example
Manage tasks across agents.
import asyncio
from agentarea.agent_kit.tasks.task_service import TaskService
from agentarea.agent_kit.agents import create_agent
async def task_example():
service = TaskService()
agent = create_agent(
name="Task Agent",
instruction="Handle assigned tasks.",
model="openai/gpt-4"
)
# Create task
task_id = await service.create_task(
description="Research AI trends",
agent=agent
)
# Monitor progress
progress = await service.get_task_progress(task_id)
print(f"Task progress: {progress}")
# Complete task
await service.complete_task(task_id, result="AI trends summary")
asyncio.run(task_example())
🌐 Agent Networks
The SDK provides AgentNetwork for building multi-agent systems:
- Event-driven communication between agents
- Role-based agent assignment
- Network orchestration for complex workflows
- Integration with tasks for distributed processing
Example: See basic network example above. For advanced setups, use network.py for custom topologies.
📋 Task Management
Robust task system via tasks/ module:
Taskcreation with descriptions, assignees, and goalsTaskServicefor CRUD operations (create, read, update, delete)- Progress evaluation with
GoalProgressEvaluator - Human-in-loop integration for oversight
- Toolset for task-related operations (e.g.,
TasksToolset)
Supports hierarchical tasks and dependencies for orchestration.
🏃 Distributed Runners
Extensible runner system for independent deployment:
BaseAgentRunner: Abstract base for custom runners- Integration points for Temporal (workflow orchestration), Restate (stateful workflows), Dapr (service invocation)
- Async execution with state persistence
- Scalable for cloud-native environments
Example integration (Temporal):
from temporalio import workflow
from agentarea.agent_kit.runners import BaseAgentRunner
class TemporalRunner(BaseAgentRunner):
@workflow.defn
async def run_workflow(self, agent, task):
# Implement Temporal workflow
return await self.execute_agent(agent, task)
Adapt for Restate or Dapr via service calls and state management.
📚 Examples
Run built-in examples:
# Network example
python examples/test_agentic_network.py
# Task example (adapt from tests)
python -m agentarea.agent_kit.example
Examples demonstrate:
- Agent networks and communication
- Task creation and assignment
- Distributed runner stubs
- Tool usage in networked contexts
See examples/ for more.
🏗️ Architecture
Modular design focused on networks, tasks, and runners:
agentarea-agents-sdk/
├── src/
│ └── agentarea.agent_kit/
│ ├── agents/ # Agents and networks
│ │ ├── agent.py # Base Agent
│ │ ├── network.py # AgentNetwork
│ │ └── multi_agent.py
│ ├── tasks/ # Task management
│ │ ├── tasks.py # Task models
│ │ └── task_service.py
│ ├── runners/ # Execution engines
│ │ └── base.py # BaseAgentRunner
│ ├── models/ # LLM integration
│ ├── tools/ # Tool system (incl. tasks_toolset)
│ ├── context/ # Context for networks
│ ├── goal/ # Task evaluation
│ └── prompts.py
├── tests/ # Tests for networks/tasks/runners
├── examples/ # Network and task examples
├── docs/
├── README.md
├── LICENSE
└── pyproject.toml
🔌 Components
Agent Networks (agents/network.py)
AgentNetwork: Orchestrate multiple agents- Event handling and inter-agent messaging
- Task routing and coordination
Task Management (tasks/)
Task: Core task entityTaskService: Service layer for operations- Integration with agents and tools
Distributed Runners (runners/)
BaseAgentRunner: For custom implementations- Support for stateful, distributed execution
Other Components
- LLM Models: Provider-agnostic
- Tools: Extensible with task-specific tools
- Prompts: ReAct for networked reasoning
📖 Supported LLM Providers
Via LiteLLM (100+ models). See examples above for usage.
OpenAI
model = LLMModel(provider_type="openai", model_name="gpt-4", api_key="your-key")
Anthropic
model = LLMModel(provider_type="anthropic", model_name="claude-3-opus-20240229", api_key="your-key")
Ollama
model = LLMModel(provider_type="ollama_chat", model_name="qwen2.5")
Full list: LiteLLM docs.
🧪 Testing
pip install -e .[dev]
pytest -q
# Network and task tests
pytest tests/test_agent.py -v
pytest tests/test_task_orchestration.py -v
pytest --cov=src/agentarea.agent_kit --cov-report=term-missing
Includes tests for networks, tasks, and runner integrations.
💻 Development
Setup:
git clone https://github.com/agentarea/agentarea-agents-sdk.git
cd agentarea-agents-sdk
python -m venv .venv
source .venv/bin/activate
pip install -e .[dev]
# Lint, type check, test
ruff check src tests
mypy src
pytest
Uses Ruff, MyPy, Pytest. Focus on network/task coverage.
🤝 Contributing
- Fork the repo
- Create feature branch:
git checkout -b feature/network-enhancement - Commit:
git commit -m 'Enhance agent networks' - Push:
git push origin feature/network-enhancement - Open PR
Emphasize contributions to networks, tasks, runners. Guidelines: PEP 8, types, tests.
See CONTRIBUTING.md for details.
📄 License
MIT License - see LICENSE.
👥 Contributors ✨
Thanks to these wonderful people:
This project follows the all-contributors specification.
Metadata
Release files for agentarea-agents-sdk 0.1.0
For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.
Source distribution (sdist)
| File | Size | Uploaded | Provenance | Actions |
|---|---|---|---|---|
| agentarea_agents_sdk-0.1.0.tar.gz | 8.0 kB | Without provenance | Details for agentarea_agents_sdk-0.1.0.tar.gz |
Built distribution (wheel)
| File | Interpreter | ABI | Platform | Actions Reset |
|---|---|---|---|---|
| agentarea_agents_sdk-0.1.0-py3-none-any.whl | Python 3 | none | any | Details for agentarea_agents_sdk-0.1.0-py3-none-any.whl |
Total release size: 78.4 kB
Release files / agentarea_agents_sdk-0.1.0.tar.gz
| Download URL | agentarea_agents_sdk-0.1.0.tar.gz |
|---|---|
| Size | 8.0 kB |
| Tags | Source |
|
SHA-256 checksum How to use checksums |
4ad31b34692f218d1fd0d8d355780f603ed742d4bd1d4e2bc25dae2ccee185e7
|
|
BLAKE2b-256 checksum How to use checksums |
4df3843e8dc8178e79eda9051c3d8e1373a0a6c1413c6a83b6f40cc1d59abb1d
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
| Uploaded via |
twine/6.2.0 CPython/3.11.12
|
Release files / agentarea_agents_sdk-0.1.0-py3-none-any.whl
| Download URL | agentarea_agents_sdk-0.1.0-py3-none-any.whl |
|---|---|
| Size | 70.4 kB |
| Tags | Python 3 |
|
SHA-256 checksum How to use checksums |
7cca52ca41e24a8d3dff14db654789a40ba81158843f49dfcc9aa54b41e74f69
|
|
BLAKE2b-256 checksum How to use checksums |
7b15364af78efb08cfb832e798fc5c4a0b5a3e095725b2c2914b043044d92280
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
| Uploaded via |
twine/6.2.0 CPython/3.11.12
|
Security for agentarea-agents-sdk 0.1.0
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
No provenance found
None of the files in this release were uploaded with provenance metadata.
A note on package security
PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.
To learn more about protecting yourself from malicious actors, visit our package security documentation.