Python SDK

The memorymachines package is the current Python SDK for the Engramme API. Use it from notebooks, scripts, backend services, and internal tools.

Installation

pip install memorymachines

The SDK source lives in MemoryMachinesInc/memorymachines_sdk_v0.

Quick Start

from memorymachines import MemoryMachines

client = MemoryMachines()

print(client.health_check())

training = client.memorize(
    file=b"Meeting notes about the Q4 roadmap and launch timeline.",
    user_name="Sanket",
    source_type="text",
    filename="meeting-notes.txt",
)

print(training["item_id"])

results = client.recall("Q4 roadmap launch timeline", source="text")

for memory in results.get("memories", []):
    print(memory)

Memorization starts asynchronous LMM training. A successful memorize response means the content was accepted and training started; the content may not be immediately recallable.

Authentication

from memorymachines import MemoryMachines

client = MemoryMachines(api_key="eng_sk_...")

from memorymachines import MemoryMachines

client = MemoryMachines()

API key resolution order:

Explicit api_key passed to MemoryMachines(...)
ENGRAMME_API_KEY
MM_API_KEY
Active profile in ~/.memorymachines/config.yaml

You can also save and manage a local key with the CLI:

memorymachines login
memorymachines whoami
memorymachines profiles
memorymachines logout

Local credentials are stored in ~/.memorymachines/config.yaml. Override that path with MM_CONFIG_PATH or ENGRAMME_CONFIG_PATH.

Methods

`MemoryMachines(...)`

client = MemoryMachines(
    api_key=None,
    profile=None,
    base_url="https://memorymachines-gateway-prod-btf57kda.uc.gateway.dev",
)

MemoryClient and Engramme are aliases for MemoryMachines.

str

API key to use directly. If omitted, the SDK checks environment variables and local profiles.

str

Local profile name from ~/.memorymachines/config.yaml.

str

API base URL.

`memorize(...)`

Submit a document for LMM training.

# From file path
result = client.memorize(
    file="notes.txt",
    user_name="Sanket",
    source_type="text",
)

# From bytes
result = client.memorize(
    file=b"Some text",
    user_name="Sanket",
    source_type="text",
    filename="note.txt",
)

str | pathlib.Path | bytes

required

File path, pathlib.Path, or raw bytes to train on.

str

Name associated with the memories.

str

default:"text"

Type of content, such as text, email, pdf, github, or google_meets.

str

Optional stable identifier for the submitted item.

str

Filename to send when file is bytes.

`recall(...)`

Retrieve memories by semantic similarity.

results = client.recall(
    "meetings about roadmap planning",
    source="text",
)

str

required

Query text, max 1,000 characters.

str

Optional source filter.

`ask(...)`

Ask for an AI-generated response based on memories.

answer = client.ask("What did I discuss about the product launch?")
print(answer["response"])

`feedback(...)`

Rate a memory for a specific query.

client.feedback(
    memory_id="item123_memory_0",
    rating=1,
    query_text="Q4 goals",
)

`health_check()`

Check API status.

ok = client.health_check()

Returns True when GET /v1/health returns HTTP 200.

Error Handling

from memorymachines import MemoryMachines, MemoryMachinesError, AuthenticationError

client = MemoryMachines()

try:
    client.recall("team meetings")
except AuthenticationError:
    print("Invalid API key")
except MemoryMachinesError as e:
    print(f"Error: {e}")

Exception	Cause
`AuthenticationError`	Invalid API key
`RateLimitError`	Too many requests
`NotFoundError`	Resource not found
`ValidationError`	Invalid parameters
`APIError`	Other API errors

Get Started

Console Setup

API Reference

SDKs

Installation

Quick Start

Authentication

Methods

`MemoryMachines(...)`

`memorize(...)`

`recall(...)`

`ask(...)`

`feedback(...)`

`health_check()`

Error Handling

Next Steps

API Reference

Authentication

​Installation

​Quick Start

​Authentication

​Methods

​MemoryMachines(...)

​memorize(...)

​recall(...)

​ask(...)

​feedback(...)

​health_check()

​Error Handling

​Next Steps

API Reference

Authentication

Installation

Quick Start

Authentication

Methods

`MemoryMachines(...)`

`memorize(...)`

`recall(...)`

`ask(...)`

`feedback(...)`

`health_check()`

Error Handling

Next Steps