Python SDK

This guide walks through the flowra Python package. Overview: SDKs.

Install

pip install flowra

From the Flowra monorepo (if unpublished):

pip install -e sdks/python

Requires Python 3.10+. The client uses the standard library only (no extra runtime deps).

Create a client

import os
from flowra import Flowra

flowra = Flowra(api_key=os.environ["FLOWRA_API_KEY"])

Optional username sets x-username for multi-tenancy:

flowra = Flowra(api_key="...", username="customer_alice")

Or clone for another end user:

alice = flowra.as_user("customer_alice")
alice.connections.create_link({"authConfigId": "..."})

List tools and run a workflow

profile = flowra.get_profile()
tools = flowra.tools.list(limit=10)

run = flowra.workflows.run(
    "WORKFLOW_ID",
    {"input": {"message": "hello from the SDK"}},
)

Skills and MCP

skills = flowra.skills.list()
servers = flowra.mcp.list()

Knowledge (per end user)

alice = flowra.as_user("customer_alice")
collections = alice.knowledge.list()

Escape hatch: `request`

Call any path on the public API:

flowra.request("GET", "/api/v1/toolkits", query={"limit": 5})

Use this when a facade method is missing. Full schemas: API reference.

Errors

HTTP failures raise SDK errors (see package flowra.errors). Catch around calls that may fail (missing connection, invalid id, …).

Getting started

Core building blocks

Data and runtime

Integrate and scale

Recipes

Native tools

Contributing

Install

Create a client

List tools and run a workflow

Skills and MCP

Knowledge (per end user)

Escape hatch: `request`

Errors

Next

​Install

​Create a client

​List tools and run a workflow

​Skills and MCP

​Knowledge (per end user)

​Escape hatch: request

​Errors

​Next

Install

Create a client

List tools and run a workflow

Skills and MCP

Knowledge (per end user)

Escape hatch: `request`

Errors

Next