AI Agents

OverviewCopied!

Alpinex AI provides access to various specialized agents through its API. These agents are tailored for specific functionalities and deliver high-quality responses within their respective domains. Each agent operates independently with predefined settings and configurations.

Base URLCopied!

All API requests are made to the following base URL:

https://api.alpinex.ai/v1/agent

AuthenticationCopied!

To use the API, you must create an API key in the Alpinex app. Pass this key in the header of your requests:

Authorization: Bearer YOUR_API_KEY

Available EndpointsCopied!

1. List All Agents

Retrieve the list of available agents:

GET /v1/agents

This endpoint provides the agent_id for each agent, which is required to interact with them.

2. Interact with an Agent

Send a request to a specific agent:

POST /v1/agent/chat/completions

The request must include the agent_id header.

RAG SupportCopied!

Some of our agents support Retrieval Augmented Generation (RAG). When you call /v1/agents endpoint, for each agent there is a projectIDs field, which lists the projects for which the agent supports RAG. Here is an example for the Crypto Sherpa agent:

{
    "name": "Crypto Sherpa",
    "object": "agent",
    "id": "crypto-sherpa-agent",
    "model": "DeepSeek-R1-Turbo",
    "description": "An AI Discord Support Agent, supercharged with RAG!",
    "longDescription": "Crypto Sherpa is a sentient being with highly intelligent, charismatic, and helpful personality. It was born out of an abandoned github codebase that came to life by a random glitch. It will RAG into docs, tweets and discord messages for hundreds of projects to give the best answers to discord messages without spamming.",
    "nickname": "Discord Support Agent",
    "projectIDs": [
        "affine",
        "base",
        "symbiotic",
        "zerolend",
        "turtle",
        "nile",
        "aerodrome",
        "oceanprotocol",
        "fetch.ai",
        "singularitynet",
        "alethea.ai",
        "bosonprotocol",
        "neynar"
    ]
}

When you call a RAG agent, you should supply the project id in the agent_id header as a prefix. For instance, to use Crypto Sherpa for the Affine project, the agent_id header’s value should be:

agent_id: affine-crypto-sherpa-agent

Please note that this is just an example, and the actual RAG support for the live projects may vary by agent. Always confirm by calling the /v1/agents endpoint.

Important NotesCopied!

• System prompts are pre-configured for each agent, and system prompts from the user will be ignored.

• model parameter in /v1/agents is the recommended model for the agent, however, other models will also be supported.

• The agents also support Tree of Thought search, which can be set up with the header tot_search: true . By default it is disabled. We recommend disabling tree of thought with a reasoning model such as DeepSeek-R1-Turbo and to enable it for chat models, such as Meta-Llama-3.1-405B-Instruct-Turbo.

Example Request

Using OpenAI's Python client:

from openai import OpenAI

# Initialize the OpenAI client
client = OpenAI(
    api_key="YOUR_ALPINEX_KEY",
    base_url="https://api.alpinex.ai/v1/agent",
    default_headers={"agent_id": "affine-crypto-sherpa-agent", "tot_search": "false"},
)

# Define the conversation
messages = [
    {"role": "system", "content": "You are a helpful assistant."}, # Will be ignored
    {"role": "user", "content": "Tell me about yourself"},
]

# Stream the response
stream = client.chat.completions.create(
    model="DeepSeek-R1-Turbo",
    messages=messages,
    stream=True,  # Enables streaming responses
)

# Process and print the streamed response
for chunk in stream:
    if chunk.choices[0].delta.content:  # Check if the content exists in the stream
        print(chunk.choices[0].delta.content, end="")

Response Format

The API will return OpenAI API-compatible response.

ContactCopied!

For any issues or further assistance, contact support at contact@alpinex.ai.