Cloud Quickstart

Prerequisites

An EverOS Cloud account
An API key: click “Create API Key”, name it, and copy it securely
Python 3.12 or higher

Before getting your API key, configure your Memory Space Scenario Mode. It influences how memories are extracted and consolidated. Learn more in the Space Scenario Mode guide.

Install the SDK

pip install everos-cloud

Create a client

from everos_cloud import EverOS

client = EverOS(api_key="your_api_key")

Never commit API keys to version control. Use environment variables in production.

Add memories

Store conversation messages into your Memory Space. Each message’s sender_id is the user the memory belongs to.

client.add(
    session_id="session_gs_001",
    messages=[
        {"sender_id": "user_demo_001", "role": "user",
         "content": "I like black Americano, no sugar, the stronger the better!"},
        {"sender_id": "user_demo_001", "role": "user",
         "content": "Today I want to discuss the project progress."},
    ],
)

The call returns status: "queued" and extraction runs in the background on its own — you do not have to trigger it. Give it a few seconds before you read the memories back.For a deterministic write — in a test, say — pass async_mode=False. The call then returns the engine’s own result ("accumulated" or "extracted") instead of an acknowledgement.

client.flush(session_id) forces extraction for a session that is still open, but it only sees messages that have already landed. Called immediately after a default (queued) add it returns "no_extraction" — nothing has arrived yet. That is not an error, and the extraction still happens on its own moments later.

Retrieve memories

Fetch a user’s stored memories by type.

data = client.get("episode", user_id="user_demo_001")
print(f"Fetched {data.count} memories")

Search memories

Find relevant memories using vector, keyword, hybrid, or agentic retrieval.

results = client.search(
    "coffee preference",
    user_id="user_demo_001",
    method="vector",
    top_k=5,
)
episodes = results.episodes or []
print(f"Found {len(episodes)} memories")

The vector index lags extraction by a few seconds. To read back something you just wrote, use client.get(...) — it reads the stored memories directly and sees them as soon as they are extracted.

What’s Next?

Core Concepts

Understand the memory lifecycle, memory types, and how EverOS processes your data.

API Reference

Explore the full API with detailed parameter docs, request/response examples, and error codes.

Knowledge Bases

Ingest documents into a searchable topic library and query it alongside memory.

Console

Monitor quota usage, track API calls, and manage keys in the EverOS Console.

Need help? Join our community.

Getting started

Core Concepts

Essential Features

Cloud Quickstart

Prerequisites

What’s Next?

Core Concepts

API Reference

Knowledge Bases

Console

​Prerequisites

​What’s Next?

Core Concepts

API Reference

Knowledge Bases

Console

Prerequisites

What’s Next?