aigents-sdk

AIgents Python Client

Official Python client library for the AIgents AI platform.

Installation

pip install aigents-sdk

Quick Start

from aigents import AIgentsClient

# Initialize the client with your API key
client = AIgentsClient(
    api_key="aig_your_api_key_here",
    base_url="http://localhost:8000"  # or your AIgents server URL
)

# List your applications
apps = client.applications.list()
print(f"Found {len(apps)} applications")

# Get a specific application
app = client.applications.get("app_id")
print(f"Application: {app.name}")

# List agents in an application
agents = client.applications.agents.list("app_id")
print(f"Found {len(agents)} agents")

# Start a chat session
session = client.chat.create_session("app_id", agent_id="agent_id")
print(f"Created session: {session.id}")

# Send a message
response = client.chat.send_message(
    session_id=session.id,
    content="Hello, how can you help me today?"
)
print(f"Assistant: {response.content}")

Advanced Usage

Tool Calls and MCP Approval

# Send a message that might trigger tool calls
response = client.chat.send_message(
    session_id=session.id,
    content="What's the weather like in San Francisco?",
    tools=[{"type": "weather_api"}]
)

# Handle tool calls
if response.tool_calls:
    for tool_call in response.tool_calls:
        print(f"Tool call: {tool_call.name}")
        print(f"Arguments: {tool_call.arguments}")

# Handle MCP approval requests
if response.requires_mcp_approval:
    print(f"MCP approval required: {response.mcp_approval_request_id}")
    
    # Approve the MCP request
    approved_response = client.chat.approve_mcp_request(
        request_id=response.mcp_approval_request_id,
        approved=True
    )

Working with Datasets

# List datasets
datasets = client.applications.datasets.list("app_id")

# Add data to a dataset
client.applications.datasets.add_data(
    app_id="app_id",
    dataset_id="dataset_id",
    data=[
        {
            "input": "What is machine learning?",
            "output": "Machine learning is a subset of AI...",
            "metadata": {"category": "education"}
        }
    ]
)

Error Handling

from aigents.exceptions import AIgentsError, AuthenticationError, RateLimitError

try:
    response = client.chat.send_message(session_id, "Hello")
except AuthenticationError:
    print("Invalid API key")
except RateLimitError:
    print("Rate limit exceeded")
except AIgentsError as e:
    print(f"API error: {e}")

Configuration

The client can be configured with environment variables:

export AIGENTS_API_KEY="aig_your_api_key_here"
export AIGENTS_BASE_URL="https://api.aigents.co"

# Client will automatically use environment variables
client = AIgentsClient()

API Reference

Client

• AIgentsClient(api_key, base_url, timeout)

Applications

• client.applications.list() - List all applications
• client.applications.get(app_id) - Get application details
• client.applications.create(name, description) - Create new application
• client.applications.update(app_id, **kwargs) - Update application
• client.applications.delete(app_id) - Delete application

Agents

• client.applications.agents.list(app_id) - List agents
• client.applications.agents.get(app_id, agent_id) - Get agent details
• client.applications.agents.create(app_id, **kwargs) - Create agent
• client.applications.agents.update(app_id, agent_id, **kwargs) - Update agent
• client.applications.agents.delete(app_id, agent_id) - Delete agent

Chat

• client.chat.create_session(app_id, **kwargs) - Create chat session
• client.chat.send_message(session_id, content, **kwargs) - Send message
• client.chat.get_history(session_id) - Get chat history
• client.chat.approve_mcp_request(request_id, approved) - Approve MCP request

Tools & MCPs

• client.applications.tools.list(app_id) - List tools
• client.applications.mcps.list(app_id) - List MCPs

Datasets

• client.applications.datasets.list(app_id) - List datasets
• client.applications.datasets.get(app_id, dataset_id) - Get dataset
• client.applications.datasets.add_data(app_id, dataset_id, data) - Add data

Contributing

1. Clone the repository
2. Install development dependencies: pip install -e .[dev]
3. Run tests: pytest
4. Format code: black aigents/
5. Type check: mypy aigents/

License

MIT License - see LICENSE file for details.