### Run Simple Example Script

Source: https://github.com/agentuniverse-ai/agentuniverse/blob/master/examples/sample_apps/is_agent_app/README.md

Navigate to the application directory and run the simple example script to start the IS agent.

```bash
cd examples/sample_apps/is_agent_app
python simple_example.py
```

--------------------------------

### Install agentUniverse and Visualization Packages

Source: https://github.com/agentuniverse-ai/agentuniverse/blob/master/docs/guidebook/en/Get_Start/2.Run_Your_First_Tutorial_Example.md

Install the core agentUniverse framework package along with the necessary visualization interface packages. This is a prerequisite for running the tutorial example.

```shell
# framework package
pip install agentUniverse

# Visualization Interface Package
pip install magent-ui ruamel.yaml
```

--------------------------------

### Start vLLM Server with Docker

Source: https://github.com/agentuniverse-ai/agentuniverse/blob/master/examples/sample_standard_app/intelligence/agentic/llm/buildin/vllm/README.md

Deploy vLLM using Docker for easier management and isolation. Ensure Docker is installed and GPU access is configured.

```bash
docker run --gpus all \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    -p 8000:8000 \
    --ipc=host \
    vllm/vllm-openai:latest \
    --model meta-llama/Llama-3.1-8B-Instruct
```

--------------------------------

### Start Basic vLLM Server

Source: https://github.com/agentuniverse-ai/agentuniverse/blob/master/examples/sample_standard_app/intelligence/agentic/llm/buildin/vllm/README.md

Start a basic vLLM server with a single model. This is suitable for development and testing.

```bash
python -m vllm.entrypoints.openai.api_server \
    --model meta-llama/Llama-3.1-8B-Instruct \
    --port 8000
```

--------------------------------

### Configuration File Example

Source: https://github.com/agentuniverse-ai/agentuniverse/blob/master/docs/guidebook/en/Get_Start/6.aU_Old_and_New_Project_Migration_Guide.md

Example of a configuration file (config.toml) showing the CORE_PACKAGE setting, which needs to be updated for package scanning in the new project structure.

```toml
CORE_PACKAGE = "intelligence"

```

--------------------------------

### Install vLLM

Source: https://github.com/agentuniverse-ai/agentuniverse/blob/master/examples/sample_standard_app/intelligence/agentic/llm/buildin/vllm/README.md

Install the vLLM library using pip. This is a prerequisite for running vLLM servers.

```bash
pip install vllm
```

--------------------------------

### Install agentUniverse

Source: https://github.com/agentuniverse-ai/agentuniverse/blob/master/examples/sample_apps/simple_qa_agent_app/README.md

Install the agentUniverse framework using pip or from source. This is a prerequisite for running the application.

```bash
pip install agentUniverse
```

```bash
git clone https://github.com/agentuniverse-ai/agentUniverse.git
cd agentUniverse
pip install -e .
```

--------------------------------

### Start MCP Server with MCPServerManager

Source: https://github.com/agentuniverse-ai/agentuniverse/blob/master/docs/guidebook/en/How-to/Use and Publish MCP Server/How_to_Publish_MCP_Servers.md

Use the `MCPServerManager` to start the MCP server, which automatically publishes tools and toolkits marked for MCP publishing. The `AgentUniverse().start(core_mode=True)` must be called before starting the server.

```python
from agentuniverse.agent_serve.web.mcp.mcp_server_manager import MCPServerManager
from agentuniverse.base.agentuniverse import AgentUniverse

class ServerApplication:
    """
    Server application.
    """

    @classmethod
    def start(cls):
        AgentUniverse().start(core_mode=True)
        MCPServerManager().start_server()
        
if __name__ == "__main__":
    ServerApplication.start()
```

--------------------------------

### Start Milvus Docker Container

Source: https://github.com/agentuniverse-ai/agentuniverse/blob/master/docs/guidebook/en/In-Depth_Guides/Tech_Capabilities/Storage/Milvus.md

Use these shell commands to download the Milvus installation script and start a standalone Docker container. This provides database services on port 19530.

```shell
curl -sfL https://raw.githubusercontent.com/milvus-io/milvus/master/scripts/standalone_embed.sh
bash standalone_embed.sh start
```

--------------------------------

### Agent Configuration Example

Source: https://github.com/agentuniverse-ai/agentuniverse/blob/master/docs/guidebook/en/In-Depth_Guides/Tutorials/Agent/Agent_Create_And_Use.md

This is an example of an agent configuration file. It shows standard configuration items and highlights the use of variables within prompts for dynamic settings.

```yaml
#    - 'knowledge_a'
metadata:
  type: 'AGENT'
  module: 'sample_standard_app.intelligence.agentic.agent.agent_instance.demo_agent'
  class: 'DemoAgent'
```

--------------------------------

### Start Simple Q&A Agent Web Server

Source: https://github.com/agentuniverse-ai/agentuniverse/blob/master/examples/sample_apps/simple_qa_agent_app/README.md

Launch the web server for the Simple Q&A Agent. This method is recommended for interacting with the agent.

```bash
cd examples/sample_apps/simple_qa_agent_app
python bootstrap/intelligence/server_application.py
```

--------------------------------

### Install FAISS CPU or GPU Version

Source: https://github.com/agentuniverse-ai/agentuniverse/blob/master/docs/guidebook/en/In-Depth_Guides/Tech_Capabilities/Storage/FAISS.md

Install the FAISS library. Use `faiss-cpu` for CPU-only or `faiss-gpu` for GPU-accelerated versions.

```bash
# For CPU-only version
pip install faiss-cpu

# For GPU-accelerated version (requires CUDA)
pip install faiss-gpu
```

--------------------------------

### Start AgentUniverse Web Server

Source: https://github.com/agentuniverse-ai/agentuniverse/blob/master/docs/guidebook/en/Get_Start/1.Application_Project_Structure_and_Explanation.md

Use this statement to start the application service. Ensure the config file path is correct if it has been changed.

```python
from agentuniverse.agent_serve.web.web_booster import start_web_server
from agentuniverse.base.agentuniverse import AgentUniverse


class ServerApplication:
    """
    Server application.
    """

    @classmethod
    def start(cls):
        AgentUniverse().start()
        start_web_server()

ServerApplication.start()

```

--------------------------------

### Start Product Service

Source: https://github.com/agentuniverse-ai/agentuniverse/blob/master/docs/guidebook/en/How-to/Guide to Visual Agentic Workflow Platform/Product_Platform_Quick_Start.md

Run the product_application.py script to initiate the agentUniverse product service. This script is located in the sample_standard_app bootstrap directory.

```python
from core.utils.common import load_yaml
from core.app.app import App

if __name__ == '__main__':
    App(load_yaml('config.yaml')).run()

```

--------------------------------

### Python Usage Example for GRR Agent

Source: https://github.com/agentuniverse-ai/agentuniverse/blob/master/examples/sample_apps/grr_agent_app/README.md

This Python code demonstrates how to initialize AgentUniverse, get an instance of the GRR agent, run it with an input prompt, and retrieve the generated output.

```python
from agentuniverse.agent.agent import Agent
from agentuniverse.agent.agent_manager import AgentManager
from agentuniverse.base.agentuniverse import AgentUniverse

# 启动 AgentUniverse
AgentUniverse().start(config_path='config/config.toml')

# 获取 GRR agent 实例
grr_agent: Agent = AgentManager().get_instance_obj('demo_grr_agent')

# 运行 agent
result = grr_agent.run(input='写一篇关于人工智能的短文')

# 获取结果
output = result.get('output')
print(output)
```

--------------------------------

### Configuring Kimi LLM via TOML File

Source: https://github.com/agentuniverse-ai/agentuniverse/blob/master/docs/guidebook/en/In-Depth_Guides/Components/LLMs/Kimi_LLM_Use.md

Example of setting Kimi LLM configuration parameters, including API key and proxy, within a TOML file. This method is an alternative to Python environment variable setup.

```toml
KIMI_API_KEY="sk-******"
KIMI_PROXY="https://xxxxxx"
KIMI_API_BASE='https://api.moonshot.cn/v1'
KIMI_ORGANIZATION = ''
```

--------------------------------

### Start MCP Server Programmatically

Source: https://github.com/agentuniverse-ai/agentuniverse/blob/master/docs/guidebook/en/In-Depth_Guides/Tech_Capabilities/Service/MCP_Server.md

This Python code demonstrates how to start the MCP server using `MCPServerManager`. Ensure `AgentUniverse` is started first. The server can be configured with host, port, and transport type (sse or streamable_http).

```python
from agentuniverse.agent_serve.web.mcp.mcp_server_manager import MCPServerManager
from agentuniverse.base.agentuniverse import AgentUniverse


class ServerApplication:
    """
    Server application.
    """

    @classmethod
    def start(cls):
        AgentUniverse().start(core_mode=True)
        MCPServerManager().start_server()
        
if __name__ == "__main__":
    ServerApplication.start()
```

--------------------------------

### Install agentUniverse from Source

Source: https://github.com/agentuniverse-ai/agentuniverse/blob/master/examples/sample_apps/simple_qa_agent_app/README.md

Clone the agentUniverse repository and install it locally. This is useful for development or when using the latest unreleased features.

```bash
git clone https://github.com/agentuniverse-ai/agentUniverse.git
cd agentUniverse
pip install -e .
```

--------------------------------

### gRPC Call Example

Source: https://github.com/agentuniverse-ai/agentuniverse/blob/master/docs/guidebook/en/In-Depth_Guides/Tech_Capabilities/Service/gRPC.md

A Python example demonstrating how to connect to the gRPC server and invoke the `service_run` and `service_run_result` methods.

```APIDOC
## gRPC Call Example

This example demonstrates how to use the gRPC client to interact with the Agent service:

```python
import grpc
from agentuniverse.agent_serve.web.rpc.grpc import agentuniverse_service_pb2, \
    agentuniverse_service_pb2_grpc


# Ensure your server_application is started first.
def test_run():
    with grpc.insecure_channel('localhost:50051') as channel:
        stub = agentuniverse_service_pb2_grpc.AgentUniverseServiceStub(channel)
        
        # Example of calling service_run
        response = stub.service_run(agentuniverse_service_pb2.AgentServiceRequest(
            service_id='demo_service',
            params='{"input":"(18+3-5)/2*4=?"}',
            saved=True
        ))
        print("client received: " + response.request_id)

        # Example of calling service_run_result
        response = stub.service_run_result(agentuniverse_service_pb2.AgentResultRequest(
            request_id=response.request_id
        ))
        print("client received: " + response.result)
```
```

--------------------------------

### Implementation Agent Configuration Example

Source: https://github.com/agentuniverse-ai/agentuniverse/blob/master/examples/sample_apps/is_agent_app/README.md

YAML configuration for the implementation agent, specifying its prompt version and LLM model settings.

```yaml
info:
  name: 'demo_implementation_agent'
  description: 'Demo implementation agent'

profile:
  prompt_version: 'demo_implementation_agent.cn'
  llm_model:
    name: 'qwen_llm'
    temperature: 0.7        # Higher creativity
```

--------------------------------

### Task Decomposition Example (Not Recommended)

Source: https://github.com/agentuniverse-ai/agentuniverse/blob/master/examples/sample_apps/is_agent_app/README.md

Illustrates an example of task decomposition with an insufficient number of checkpoints, highlighting a suboptimal approach for complex tasks.

```python
# Not recommended: too few checkpoints
result = agent.run(
    input='开发完整的Web应用',
    checkpoint_count=2  # Too few
)
```

--------------------------------

### Gunicorn Configuration File Example

Source: https://github.com/agentuniverse-ai/agentuniverse/blob/master/docs/guidebook/en/In-Depth_Guides/Tech_Capabilities/Service/Web_Server.md

An example TOML file for configuring Gunicorn server settings. This file allows detailed customization of worker processes, threads, timeouts, and more.

```toml
[GUNICORN_CONFIG]
bind = '127.0.0.1:8888'
backlog = 2048
worker_class = 'gthread'
threads = 4
workers = 5
timeout = 60
keepalive = 10
```

--------------------------------

### SQLDBWrapper Configuration File Example

Source: https://github.com/agentuniverse-ai/agentuniverse/blob/master/docs/guidebook/en/In-Depth_Guides/Tech_Capabilities/Storage/SQLDB_WRAPPER.md

An example of a YAML configuration file for registering an SQLDBWrapper instance. This includes details like name, database URI, and optional arguments for SQLDatabase and SQLAlchemy engine.

```APIDOC
## SQLDBWrapper Configuration File

```yaml
name: 'demo_sql'
description: 'demo_sql'
db_uri: "sqlite:///./demo.db"
sql_database_args:
  include_tables: ["users"]
engine_args:
  pool_size: 10
metadata:
  type: 'SQLDB_WRAPPER'
```

- **`name`**: The name of the SQLDBWrapper, used to uniquely identify an instance.
- **`description`**: A brief description of the functionality of the SQLDBWrapper's.
- **`db_uri`**: A SQLAlchemy-style database URI, which is used to create a SQLAlchemy engine.
- **`sql_database_args`**: Optional parameters for the SQLDatabase class in `LangChain`. For specific details, refer to [LangChain's official website](https://python.langchain.com/v0.1/docs/integrations/toolkits/sql_database/).
- **`engine_args`**:  Optional parameters used to configure SQLAlchemy engine settings. For specific configurable options, please refer to [SQLAlchemy's webpage](https://docs.sqlalchemy.org/en/20/core/engines.html#sqlalchemy.create_engine).
- **`metadata`**: Indicates that this configuration is specific to the `SQLDB_WRAPPER`and should not be changed.
```

--------------------------------

### Documentation Writing Task Example

Source: https://github.com/agentuniverse-ai/agentuniverse/blob/master/examples/sample_apps/is_agent_app/README.md

Example of using the IS agent for writing technical documentation, with specified checkpoints and corrections to ensure completeness and accuracy.

```python
result = agent.run(
    input='编写产品使用手册,包括安装、配置和常见问题',
    checkpoint_count=3,
    max_corrections=2
)
```

--------------------------------

### Configure and Start gRPC Server

Source: https://github.com/agentuniverse-ai/agentuniverse/blob/master/docs/guidebook/en/In-Depth_Guides/Tech_Capabilities/Service/gRPC.md

Enable the gRPC server by setting 'activate' to 'true' in the config.toml file. Then, start the server using the provided Python script. The 'max_workers' and 'server_port' parameters can also be configured.

```toml
[GRPC]
activate = 'true'
max_workers = 10
server_port = 50051
```

--------------------------------

### Verify agentUniverse Installation

Source: https://github.com/agentuniverse-ai/agentuniverse/blob/master/examples/sample_apps/simple_qa_agent_app/README.md

Run the provided test script to confirm that the agentUniverse installation and the simple Q&A agent are set up correctly.

```bash
cd examples/sample_apps/simple_qa_agent_app
python intelligence/test/simple_qa_agent_test.py
```

--------------------------------

### Get and Run a Tool Instance

Source: https://github.com/agentuniverse-ai/agentuniverse/blob/master/docs/guidebook/en/In-Depth_Guides/Tutorials/Tool/Tool_Create_And_Use.md

Obtain a tool instance by its name from the ToolManager and execute it with provided inputs.

```python
from agentuniverse.agent.action.tool.tool_manager import ToolManager

tool = ToolManager().get_instance_obj('your_tool_name')
tool_input = {'your_input_key': 'input_values'}
tool.run(**tool_input)
```

--------------------------------

### IS Agent Configuration Example

Source: https://github.com/agentuniverse-ai/agentuniverse/blob/master/examples/sample_apps/is_agent_app/README.md

YAML configuration for the main IS agent, defining its sub-agents, default checkpoint count, and maximum corrections.

```yaml
info:
  name: 'demo_is_agent'
  description: 'Demo IS agent for supervised task execution'

profile:
  # Sub-agent configuration
  implementation: 'demo_implementation_agent'
  supervision: 'demo_supervision_agent'

  # Checkpoint configuration
  checkpoint_count: 3        # Default checkpoint count
  max_corrections: 2         # Maximum correction times

  # Expert framework (optional)
  expert_framework_enabled: false
```

--------------------------------

### Direct API Key Configuration

Source: https://github.com/agentuniverse-ai/agentuniverse/blob/master/docs/guidebook/en/In-Depth_Guides/Components/LLMs/BaiChuan_LLM_Use.md

Example of setting the API key directly as a string within the configuration file.

```yaml
api_key: 'sk-xxx'
```

--------------------------------

### Configure WenXin API Keys via TOML

Source: https://github.com/agentuniverse-ai/agentuniverse/blob/master/docs/guidebook/en/In-Depth_Guides/Components/LLMs/WenXin_LLM_Use.md

Add QIANFAN_AK and QIANFAN_SK configurations to a custom_key.toml file for environment variable setup.

```toml
QIANFAN_AK="sk-******"
QIANFAN_SK="https://xxxxxx"
```

--------------------------------

### Build Project Image

Source: https://github.com/agentuniverse-ai/agentuniverse/blob/master/docs/guidebook/en/In-Depth_Guides/Deployment/Project_Image_Packaging.md

Navigate to the image build directory and execute the build script. Replace 'sample_standard_app' with your project's name.

```shell
cd xxx/xxx/sample_standard_app/image_build

chmod +x start_build.sh
./start_build.sh
```

--------------------------------

### Start agentUniverse Framework

Source: https://context7.com/agentuniverse-ai/agentuniverse/llms.txt

Initializes the framework by loading configurations, registering components, and setting up logging and tracing. Use `start_web_server()` to launch the API server after initialization.

```python
from agentuniverse.base.agentuniverse import AgentUniverse
from agentuniverse.agent_serve.web.web_booster import start_web_server

# Minimal startup — reads config from <project_root>/config/config.toml by default
AgentUniverse().start()

# With explicit config path
AgentUniverse().start(config_path="/app/config/config.toml")

# config/config.toml structure
# [BASE_INFO]
# appname = 'my_agent_app'
# [CORE_PACKAGE]
# default = ['my_app.intelligence.agentic']
# [SUB_CONFIG_PATH]
# custom_key_path = './custom_key.toml'
# log_config_path = './logging.toml'

# After start(), launch the web server
start_web_server()
```

--------------------------------

### Get Work Pattern Instance

Source: https://github.com/agentuniverse-ai/agentuniverse/blob/master/docs/guidebook/en/In-Depth_Guides/Tutorials/WorkPattern/WorkPattern.md

Use the `get_instance_obj` method from `WorkPatternManager` to retrieve a work pattern instance by its name. Ensure the `agentuniverse` library is installed and the work pattern is configured.

```python
from agentuniverse.agent.work_pattern.work_pattern_manager import WorkPatternManager

work_pattern_name = 'xxx_work_pattern'
work_pattern = WorkPatternManager().get_instance_obj(component_instance_name=work_pattern_name)
```

--------------------------------

### AgentUniverse.start()

Source: https://context7.com/agentuniverse-ai/agentuniverse/llms.txt

Initializes the agentUniverse framework by loading configurations, setting up logging and tracing, and registering all defined components. It can be called with a default or explicit configuration path.

```APIDOC
## AgentUniverse.start()

### Description
Initializes the entire agentUniverse framework. This includes loading configuration files, setting up logging and tracing, scanning for component definitions (agents, LLMs, tools, etc.), and registering them. It serves as the bootstrap function for the framework.

### Method
```python
AgentUniverse().start(config_path: str = None)
```

### Parameters
*   **config_path** (str, Optional) - The explicit path to the configuration file. If not provided, it defaults to `<project_root>/config/config.toml`.

### Request Example
```python
from agentuniverse.base.agentuniverse import AgentUniverse

# Minimal startup
AgentUniverse().start()

# With explicit config path
AgentUniverse().start(config_path="/app/config/config.toml")
```

### Notes
After calling `start()`, you typically proceed to launch other services, such as the web server using `start_web_server()`.
```

--------------------------------

### Verify agentUniverse installation

Source: https://github.com/agentuniverse-ai/agentuniverse/blob/master/docs/guidebook/en/Get_Start/0.Installation_and_Initialization.md

Check if agentUniverse is installed by listing installed packages and filtering for 'agentUniverse'. A successful installation will display the package name and its version.

```shell
pip list | grep agentUniverse
```

--------------------------------

### HTTP GET Tool

Source: https://github.com/agentuniverse-ai/agentuniverse/blob/master/docs/guidebook/en/In-Depth_Guides/Components/Tools/Integrated_Tools.md

The HTTP GET tool sends a GET request to a specified URL. The input should be a URL, and the output will be the text response of the GET request.

```APIDOC
## HTTP GET Tool

### Description
This tool can send a GET request. Input should be a URL (e.g., https://www.google.com). The output will be the text response of the GET request.

### Tool Configuration
```yaml
name: 'requests_get'
description: 'A portal to the internet. Use this when you need to get specific
    content from a website. Input should be a  url (i.e. https://www.google.com).
    The output will be the text response of the GET request.
        ```'
headers:
  content-type: 'application/json'
method: 'GET'
json_parser: false
response_content_type: json
tool_type: 'api'
input_keys: ['input']
metadata:
  type: 'TOOL'
  module: 'sample_standard_app.intelligence.agentic.tool.request_tool'
  class: 'RequestTool'
```

### Usage
This tool can be used directly without requiring any keys.
```

--------------------------------

### Start vLLM Server with Quantization

Source: https://github.com/agentuniverse-ai/agentuniverse/blob/master/examples/sample_standard_app/intelligence/agentic/llm/buildin/vllm/README.md

Launch a vLLM server with quantization enabled (e.g., AWQ) to reduce memory footprint and potentially improve inference speed.

```bash
python -m vllm.entrypoints.openai.api_server \
    --model TheBloke/Llama-2-70B-Chat-AWQ \
    --quantization awq \
    --port 8000
```

--------------------------------

### Agent Target Examples

Source: https://github.com/agentuniverse-ai/agentuniverse/blob/master/docs/guidebook/en/In-Depth_Guides/Tutorials/Agent/Agent.md

Provides examples of good and bad agent target descriptions. Good examples are concise, impactful, and specify a domain scope.

```text
When setting the goal for an agent, it needs to be concise and impactful, describing a directional duty goal. Providing a specific and specialized domain scope is optimal.

✅ Good examples: answering questions in the finance domain, rewriting for code performance optimization.

❌ Bad examples: aimless descriptions such as "just chatting."
```

--------------------------------

### Install AgentUniverse with SLS Support

Source: https://github.com/agentuniverse-ai/agentuniverse/blob/master/docs/guidebook/en/In-Depth_Guides/Tech_Capabilities/Log_And_Monitor/Alibaba_Cloud_SLS.md

Install the AgentUniverse package with the log extension to enable SLS integration. This command installs the necessary dependencies for log management capabilities.

```shell
pip install agentuniverse[log_ext]
```

--------------------------------

### Startup Configuration with Threads

Source: https://github.com/agentuniverse-ai/agentuniverse/blob/master/docs/guidebook/en/In-Depth_Guides/Tech_Capabilities/Service/Web_Server.md

Configures the web server at startup, allowing specific parameters like bind address and number of threads to be set directly. This overrides settings from configuration files.

```python
from agentuniverse.agent_serve.web.web_booster import start_web_server
from agentuniverse.base.agentuniverse import AgentUniverse

AgentUniverse().start()
start_web_server(bind="127.0.0.1:8002", threads=4)
```

--------------------------------

### LLM Configuration Example (Qwen)

Source: https://github.com/agentuniverse-ai/agentuniverse/blob/master/examples/sample_apps/grr_agent_app/README.md

Configuration for the Qwen LLM, specifying model name, maximum tokens, and temperature. It also includes metadata for the DefaultOpenAILLM class.

```yaml
name: 'qwen_llm'
description: 'Qwen LLM configuration'
model_name: 'qwen-max'
max_tokens: 4096
temperature: 0.7
metadata:
  type: 'LLM'
  module: 'agentuniverse.llm.default.default_openai_llm'
  class: 'DefaultOpenAILLM'
```

--------------------------------

### Direct and Async Tool Usage

Source: https://context7.com/agentuniverse-ai/agentuniverse/llms.txt

Demonstrates how to instantiate and run a tool directly or asynchronously using the ToolManager. Requires the tool to be registered.

```python
# Direct usage
from agentuniverse.agent.action.tool.tool_manager import ToolManager

tool = ToolManager().get_instance_obj('stock_price_tool')
result = tool.run(ticker='AAPL')
print(result)  # AAPL current price: $213.49

# Async usage
import asyncio
result = asyncio.run(tool.async_run(ticker='GOOG'))
print(result)  # GOOG current price: $175.12
```

--------------------------------

### Install AgentUniverse Dependency

Source: https://github.com/agentuniverse-ai/agentuniverse/blob/master/examples/sample_apps/is_agent_app/README.md

Install the agentuniverse library using pip. This is a prerequisite for running the application.

```bash
pip install agentuniverse
```

--------------------------------

### Install Milvus Python SDK

Source: https://github.com/agentuniverse-ai/agentuniverse/blob/master/docs/guidebook/en/In-Depth_Guides/Tech_Capabilities/Storage/Milvus.md

Install the necessary Python SDK to interact with Milvus from your agentUniverse project.

```python
pip install pymilvus
```

--------------------------------

### Chroma Store Configuration Example

Source: https://github.com/agentuniverse-ai/agentuniverse/blob/master/docs/guidebook/en/How-to/Build and Use a Knowledge Base/How_to_Build_Knowledge_Base.md

Configuration for a ChromaDB store, including its name, description, persistence path, embedding model, and similarity search parameters.

```yaml
name: 'civil_law_chroma_store'
description: '保存了中国民法典的所有内容，以文本向量形式存储'
persist_path: '../../db/civil_law.db'
embedding_model: 'dashscope_embedding'
similarity_top_k: 100
metadata:
  type: 'STORE'
  module: 'agentuniverse.agent.action.knowledge.store.chroma_store'
  class: 'ChromaStore'
```

--------------------------------

### Starting the gRPC Server

Source: https://github.com/agentuniverse-ai/agentuniverse/blob/master/docs/guidebook/en/In-Depth_Guides/Tech_Capabilities/Service/gRPC.md

Python code to start the AgentUniverse application and the web server, which includes the gRPC service if activated.

```APIDOC
## Starting the gRPC Server

After configuring `config.toml`, start the gRPC server using the following Python code:

```python
from agentuniverse.agent_serve.web.web_booster import start_web_server
from agentuniverse.base.agentuniverse import AgentUniverse

AgentUniverse().start()
start_web_server()
```
```

--------------------------------

### Install agentUniverse via pip

Source: https://github.com/agentuniverse-ai/agentuniverse/blob/master/docs/guidebook/en/Get_Start/0.Installation_and_Initialization.md

Use this command to install the agentUniverse package using pip. Ensure you have Python 3.10 or later.

```shell
pip install agentUniverse
```

--------------------------------

### HTTP GET Tool Configuration

Source: https://github.com/agentuniverse-ai/agentuniverse/blob/master/docs/guidebook/en/In-Depth_Guides/Components/Tools/Integrated_Tools.md

Configures a tool for sending HTTP GET requests. Input should be a URL, and the output is the text response.

```yaml
name: 'requests_get'
description: 'A portal to the internet. Use this when you need to get specific
    content from a website. Input should be a  url (i.e. https://www.google.com).
    The output will be the text response of the GET request.
        ```'
headers:
  content-type: 'application/json'
method: 'GET'
json_parser: false
response_content_type: json
tool_type: 'api'
input_keys: ['input']
metadata:
  type: 'TOOL'
  module: 'sample_standard_app.intelligence.agentic.tool.request_tool'
  class: 'RequestTool'
```

--------------------------------

### Basic FAISS Store Initialization and Usage

Source: https://github.com/agentuniverse-ai/agentuniverse/blob/master/docs/guidebook/en/In-Depth_Guides/Tech_Capabilities/Storage/FAISS.md

Initialize a FAISS store with IndexFlatL2, add documents, and perform a similarity search query. Ensure the client is initialized using `_new_client()` before operations.

```python
from agentuniverse.agent.action.knowledge.store.faiss_store import FAISSStore
from agentuniverse.agent.action.knowledge.store.document import Document
from agentuniverse.agent.action.knowledge.store.query import Query

# Initialize store
store = FAISSStore(
    index_path="./data/my_faiss.index",
    metadata_path="./data/my_faiss_metadata.pkl",
    embedding_model="your_embedding_model",
    index_config={
        "index_type": "IndexFlatL2",
        "dimension": 768
    }
)

# Initialize the client
store._new_client()

# Insert documents
documents = [
    Document(text="Python is a programming language", metadata={"topic": "programming"}),
    Document(text="Machine learning uses algorithms", metadata={"topic": "AI"}),
]
store.insert_document(documents)

# Query documents
query = Query(query_str="programming languages")
results = store.query(query)

for doc in results:
    print(f"Text: {doc.text}")
    print(f"Score: {doc.metadata.get('score')}")
```

--------------------------------

### Data Processing Task Example

Source: https://github.com/agentuniverse-ai/agentuniverse/blob/master/examples/sample_apps/is_agent_app/README.md

Example of using the IS agent for a data processing task, such as designing and implementing a data cleaning pipeline.

```python
result = agent.run(
    input='设计并实现一个数据清洗流程,处理CSV文件中的缺失值和异常值',
    checkpoint_count=4,
    max_corrections=2
)
```

--------------------------------

### Initialize HumanInputRun Tool with Configuration

Source: https://github.com/agentuniverse-ai/agentuniverse/blob/master/docs/guidebook/en/In-Depth_Guides/Components/Tools/Integrated_LangChain_Tools.md

Use this configuration to initialize the HumanInputRun tool from LangChain. Set description to empty if you want to use the description directly from LangChain.

```yaml
name: 'human_input_run'
description: ''
tool_type: 'api'
input_keys: ['input']
langchain:
  module: langchain_community.tools
  class_name: HumanInputRun
metadata:
  type: 'TOOL'
  module: 'sample_standard_app.intelligence.agentic.tool.samples.langchain_tool'
  class: 'LangChainTool'
```

--------------------------------

### Code Development Task Example

Source: https://github.com/agentuniverse-ai/agentuniverse/blob/master/examples/sample_apps/is_agent_app/README.md

Example of using the IS agent for a code development task, specifying checkpoints and corrections for quality assurance.

```python
result = agent.run(
    input='实现一个RESTful API端点,包括参数验证、错误处理和文档',
    checkpoint_count=4,
    max_corrections=2
)
```

--------------------------------

### Deploy Signoz Locally

Source: https://github.com/agentuniverse-ai/agentuniverse/blob/master/docs/guidebook/en/In-Depth_Guides/Tech_Capabilities/Observability/Observability_with_OTEL.md

Use this script to clone the Signoz repository, navigate to the deployment directory, make the install script executable, and then run it to deploy Signoz locally.

```shell
git clone https://github.com/SigNoz/signoz.git
cd signoz/deploy/
chmod +x install.sh
./install.sh
```

--------------------------------

### Install agentUniverse UI dependencies

Source: https://github.com/agentuniverse-ai/agentuniverse/blob/master/README.md

Install the necessary packages for the visual agentic workflow platform. This includes the UI components and YAML handling.

```shell
pip install magent-ui ruamel.yaml
```

--------------------------------

### Python Usage of vLLM with AgentUniverse

Source: https://github.com/agentuniverse-ai/agentuniverse/blob/master/examples/sample_standard_app/intelligence/agentic/llm/buildin/vllm/README.md

Demonstrates how to load a vLLM instance using LLMManager and perform simple and streaming calls.

```python
from agentuniverse.llm.llm_manager import LLMManager

# Load vLLM instance
llm = LLMManager().get_instance_obj("vllm-llama-3.1-8b")

# Simple call
messages = [{"role": "user", "content": "Hello!"}]
response = llm.call(messages)
print(response.text)

# Streaming call
for chunk in llm.call(messages, streaming=True):
    print(chunk.text, end="", flush=True)
```

--------------------------------

### Initialize and Use Basic Document Classifier

Source: https://github.com/agentuniverse-ai/agentuniverse/blob/master/examples/third_party_examples/tools/document_classifier_tool/docs/DocumentClassifier.md

Demonstrates how to create a DocumentClassifier instance, set categories, add keyword rules, and classify a document. Ensure the Document class and DocumentClassifier are imported.

```python
from agentuniverse.agent.action.knowledge.store.document import Document
from examples.third_party_examples.tools.document_classifier_tool.document_classifier import DocumentClassifier

# 创建分类器
classifier = DocumentClassifier()

# 设置分类类别
classifier.set_categories(['技术文档', '学术论文', '新闻资讯', '其他'])

# 添加关键词规则
classifier.add_keyword_rule('技术文档', ['代码', '编程', 'API', '函数'])

# 分类文档
doc = Document(text="这是一个Python编程教程")
classified_doc = classifier._classify_document(doc)
```

--------------------------------

### Configure LLM API Key

Source: https://github.com/agentuniverse-ai/agentuniverse/blob/master/examples/sample_apps/is_agent_app/README.md

Set your LLM API key as an environment variable. This example uses Tongyi Qianwen (通义千问) as an example.

```bash
export DASHSCOPE_API_KEY="your_api_key_here"
```

--------------------------------

### Configuring PromptToolkit

Source: https://github.com/agentuniverse-ai/agentuniverse/blob/master/examples/third_party_examples/apps/prompt_toolkit_app/prompt/README.md

Sets up configuration for PromptToolkit, including enabling auto-optimization, defining default scenario and complexity, specifying optimization strategies, and setting a confidence threshold. This allows customization of the toolkit's behavior.

```python
from examples.third_party_examples.apps.prompt_toolkit_app.prompt.prompt_toolkit import PromptToolkitConfig
from examples.third_party_examples.apps.prompt_toolkit_app.prompt.prompt_generator import PromptScenario, PromptComplexity
from examples.third_party_examples.apps.prompt_toolkit_app.prompt.prompt_optimizer import OptimizationStrategy

config = PromptToolkitConfig(
    enable_auto_optimization=True,  # Enable automatic optimization
    default_scenario=PromptScenario.CONVERSATIONAL,  # Default scenario
    default_complexity=PromptComplexity.MEDIUM,  # Default complexity
    optimization_strategies=[
        OptimizationStrategy.CLARITY,
        OptimizationStrategy.STRUCTURE
    ],
    confidence_threshold=0.6  # Confidence threshold
)
```

--------------------------------

### Test Demo Agent using AgentManager

Source: https://github.com/agentuniverse-ai/agentuniverse/blob/master/docs/guidebook/en/In-Depth_Guides/Tutorials/Agent/Agent_Create_And_Use.md

Python code demonstrating how to initialize AgentUniverse, retrieve an agent instance by name from the AgentManager, and run it with input.

```python
import unittest

from agentuniverse.agent.agent import Agent
from agentuniverse.agent.agent_manager import AgentManager
from agentuniverse.agent.output_object import OutputObject
from agentuniverse.base.agentuniverse import AgentUniverse


class DemoAgentTest(unittest.TestCase):
    """
    Test cases for the demo agent
    """

    def setUp(self) -> None:
        AgentUniverse().start(config_path='../../config/config.toml')

    def test_demo_agent(self):
        """Test demo agent."""
        instance: Agent = AgentManager().get_instance_obj('demo_agent')
        output_object: OutputObject = instance.run(input="What is the reason for the sharp rise in Nvidia's stock?")
        print(output_object.get_data('output'))

if __name__ == '__main__':
    unittest.main()
```