arango-ai-sdk

arango-ai-sdk is the Python SDK for the Arango AI Suite. It is the code-first way to drive an AutoGraph workflow: authenticate, create a project, store the API keys it needs, upload documents, build the context graph, and ask questions against it. You never assemble a URL, set an authentication header, poll a job by hand, or interpret an HTTP status code. The SDK does that and raises a named exception when something goes wrong.

The SDK is organised in areas. Client-wide areas are namespaces on the client, such as client.project and client.secret_profile. Project-scoped areas are objects on the project handle those namespaces return, such as sales.autograph and sales.autorag:

Area

What it covers

Status

Authentication

Sign in once with a username and password; every later call is authenticated for you.

Available

Projects (client.project)

The project everything else happens in.

Available: create

Secret profiles (client.secret_profile)

Store a provider API key once; later stages refer to it by ID.

Available: create, get

Files (client.files)

Upload the documents to ingest.

Available: upload one file, several files, or a folder

AutoGraph (sales.autograph)

The project’s AutoGraph service: deploy it, then build the context graph from the uploaded documents with it.

Available: deploy, undeploy, ingest and its steps (corpus build, strategize, read strategies, orchestrate)

AutoRAG (sales.autorag)

Deploy a retriever and ask questions against the built context graph.

Available: deploy, list, get, ask, undeploy

Note

The SDK is pre-alpha and grows one area at a time. This documentation covers what is available today.