Obot Python Client

A Python client for interacting with the Obot API.

Installation (in development, not published to PyPI)

poetry install

The following won't work until published to PyPI:

pip install obot-python

Quick Start

from obot.sync_client import ObotClient

client = ObotClient(
    base_url="http://localhost:8080",
    token="your-token"
)

Chat

The chat interface provides a simple way to interact with agents:

# Start a conversation
conv = client.chat("a18pjdh", "Hi there!")
print(conv)  # prints the agent's response

# Continue the conversation using the same thread
response = conv.chat("What's 2+2?")
print(response)

# Stream responses in real-time
for chunk in conv.chat("Tell me a story", stream=True):
    print(chunk, end="", flush=True)

Async Support

For async applications:

from obot.async_client import AsyncObotClient
import asyncio

async def main():
    client = AsyncObotClient(
        base_url="http://localhost:8080",
        token="your-token"
    )
    
    # Start a conversation
    conv = await client.chat("a18pjdh", "Hi there!")
    print(conv)  # prints response
    
    # Continue conversation
    response = await conv.achat("What's 2+2?")
    print(response)
    
    # Stream responses
    async for chunk in await conv.achat("Tell me a story", stream=True):
        print(chunk, end="", flush=True)

if __name__ == "__main__":
    asyncio.run(main())

Tools

List all available tools and their categories:

# List all tools
all_tools = client.tools()

# List all tool categories
categories = client.tools.categories()

# List tools in a specific category
slack_tools = client.tools(category="Slack")

Models

Work with models and model providers:

# List all model providers
providers = client.models.providers()

# List only configured providers
configured_providers = client.models.providers(configured=True)

# List all models
all_models = client.models()

# List active models from a specific provider
active_models = client.models(
    model_provider="openai-model-provider",
    active=True
)

Agents

Create and manage agents:

# List all agents
all_agents = client.agents()

# Get a specific agent
agent = client.agents.get("a1-obot")

# Create a new agent
new_agent = client.agents.create(
    name="Test Agent",
    description="A test agent",
    prompt="You are a helpful assistant",
    model="grok-beta",  # Will be converted to model ID
    tools=["google-search-bundle", "slack-send-message"]  # Tool IDs
)

# Update specific fields of an agent
updated_agent = client.agents.update(
    agent_id="a1-agent-id",
    name="New Name",  # Only updates the name
    tools=["google-search-bundle"]  # Only updates the tools
)

Notes

• When creating or updating agents, model names are automatically converted to model IDs
• Tool IDs are validated to ensure they exist
• Updates only modify the specified fields, preserving other settings
• Chat responses can be streamed in real-time
• Both synchronous and asynchronous clients are supported
• Conversations maintain thread context automatically

TODO

• Workflows API
• Credentials API
• Webhooks API
• Error handling documentation
• Rate limiting
• Pagination support
• OAuth support
• CLI tool
• Type hints documentation
• More examples and use cases
• Unit tests
• Integration tests

Development

To contribute to this project:

1. Clone the repository
2. Install Poetry (if you haven't already): curl -sSL https://install.python-poetry.org | python3 -
3. Install dependencies: poetry install

You can also activate the virtual environment directly:

poetry shell
pytest