AI Automation Agent with Manus.im-like capabilities for full task automation
A comprehensive open-source AI automation agent that provides Manus AI-like capabilities for full task automation. Unlike traditional AI assistants that only provide suggestions, this agent autonomously executes tasks across multiple domains including research, content creation, software development, data analysis, and workflow automation.
Like Manus AI, goes beyond suggestions to actually execute tasks
Process and generate multiple types of data like Manus AI
Seamless integration with external applications
Designed to excel in real-world task automation benchmarks
Compete with commercial tools while staying open
git clone https://github.com/4hid/automation-agent-manus-like.git
cd automation-agent-manus-like
pip install -r requirements.txt
cp config.example.yaml config.yaml
# Edit config.yaml with your settings
from automation_agent import AutomationAgent, Task, TaskType
# Initialize the agent
agent = AutomationAgent()
# Create a research task
task = Task(
id="research_ai_trends",
description="Research latest AI automation trends and create a comprehensive report",
task_type=TaskType.RESEARCH,
priority=1
)
# Add task to agent
agent.add_task(task)
# Execute tasks autonomously
result = agent.execute_autonomous()
print(f"Task completed: {result}")
# Create a complex multi-step project
tasks = [
Task("research", "Research market trends in AI automation", TaskType.RESEARCH),
Task("analysis", "Analyze competitor products and features", TaskType.DATA_ANALYSIS),
Task("content", "Create marketing presentation", TaskType.CONTENT_CREATION),
Task("development", "Build prototype application", TaskType.SOFTWARE_DEVELOPMENT)
]
# Add dependencies
tasks[1].dependencies = ["research"] # Analysis depends on research
tasks[2].dependencies = ["research", "analysis"] # Content depends on both
tasks[3].dependencies = ["analysis"] # Development depends on analysis
# Execute workflow
for task in tasks:
agent.add_task(task)
result = agent.execute_autonomous(max_iterations=50)
The agent can be configured through config.yaml:
agent:
autonomous_mode: true
max_iterations: 30
error_recovery: true
tools:
research:
max_sources: 10
fact_check: true
content:
output_format: "markdown"
include_citations: true
development:
test_coverage: 80
code_quality_check: true
quality_assurance:
verification_enabled: true
validation_threshold: 0.8
class AutomationAgent:
def __init__(self, config: AgentConfig = None)
def add_task(self, task: Task) -> None
def execute_autonomous(self, max_iterations: int = 30) -> dict
def get_task_status(self) -> dict
def pause_execution(self) -> None
def resume_execution(self) -> None
class Task:
def __init__(self, id: str, description: str, task_type: TaskType, priority: int = 1)
def add_dependency(self, task_id: str) -> None
def set_status(self, status: TaskStatus) -> None
def get_progress(self) -> float
from automation_agent.tools import BaseTool
class CustomTool(BaseTool):
def __init__(self):
super().__init__("custom_tool", "Custom automation tool")
def execute(self, task: Task, context: dict) -> dict:
# Implement your custom logic
return {"status": "success", "result": "Custom result"}
# Register the tool
agent.register_tool(CustomTool())
from automation_agent import TaskType
# Extend TaskType enum
class CustomTaskType(TaskType):
CUSTOM_AUTOMATION = "custom_automation"
SPECIALIZED_TASK = "specialized_task"
# Implement custom execution logic
def execute_custom_task(self, task: Task, state: State) -> Action:
# Custom task execution logic
pass
Run the test suite:
python -m pytest tests/
Run the demo:
python demo_automation_agent.py
Inspired by Manus AI’s SOTA performance in real-world task automation
| undefinedModelundefined | undefinedGAIA Accuracyundefined | undefinedKey Strengthsundefined | undefinedAvailabilityundefined |
|---|---|---|---|
| undefinedManus AIundefined | >65% (SOTA) | Autonomous execution, multi-modal, tool integration | Commercial |
| undefinedOur Agentundefined | undefinedTarget: 65%+undefined | Open-source, customizable, self-hosted | undefinedOpen Sourceundefined |
| H2O.ai (h2oGPTe) | 65% | Enterprise-grade AI, tool-enhanced | Commercial |
| Google (Langfun) | 49% | Advanced reasoning, limited tools | Limited |
| OpenAI (GPT-4o) | 32% | Plugin-based functionality | Commercial |
| undefinedFeatureundefined | undefinedManus AIundefined | undefinedOur Implementationundefined | undefinedStatusundefined |
|---|---|---|---|
| Autonomous Task Execution | ✅ | ✅ | undefinedCompleteundefined |
| Multi-Modal Capabilities | ✅ | ✅ | undefinedCompleteundefined |
| Advanced Tool Integration | ✅ | ✅ | undefinedCompleteundefined |
| Asynchronous Processing | ✅ | 🔄 | undefinedIn Progressundefined |
| Adaptive Learning | ✅ | 🔄 | undefinedPlannedundefined |
| Real-time Web Access | ✅ | ✅ | undefinedCompleteundefined |
| Code Generation & Execution | ✅ | ✅ | undefinedCompleteundefined |
| Report & Document Creation | ✅ | ✅ | undefinedCompleteundefined |
We welcome contributions! Please see CONTRIBUTING.md for guidelines.
git checkout -b feature/amazing-featurepython -m pytestgit commit -m 'Add amazing feature'git push origin feature/amazing-featureThis project is licensed under the MIT License - see the LICENSE file for details.
undefinedMade with ❤️ for the automation communityundefined