pykiwiai

A Python client library for accessing the Runkai AI API

Usage

First, create a client:

from pykiwiai import Client

client = Client()  # defaults to https://api.meetkiwi.ai

If the endpoints you're going to hit require authentication, use AuthenticatedClient instead:

from pykiwiai import AuthenticatedClient

client = AuthenticatedClient(token="SuperSecretToken")  # defaults to https://api.meetkiwi.ai

Now call your endpoint and use your models:

from pykiwiai.models import MyDataModel
from pykiwiai.api.my_tag import get_my_data_model
from pykiwiai.types import Response

with client as client:
    my_data: MyDataModel = get_my_data_model.sync(client=client)
    # or if you need more info (e.g. status_code)
    response: Response[MyDataModel] = get_my_data_model.sync_detailed(client=client)

Or do the same thing with an async version:

from pykiwiai.models import MyDataModel
from pykiwiai.api.my_tag import get_my_data_model
from pykiwiai.types import Response

async with client as client:
    my_data: MyDataModel = await get_my_data_model.asyncio(client=client)
    response: Response[MyDataModel] = await get_my_data_model.asyncio_detailed(client=client)

By default, when you're calling an HTTPS API it will attempt to verify that SSL is working correctly. Using certificate verification is highly recommended most of the time, but sometimes you may need to authenticate to a server (especially an internal server) using a custom certificate bundle.

client = AuthenticatedClient(
    base_url="https://internal_api.example.com", 
    token="SuperSecretToken",
    verify_ssl="/path/to/certificate_bundle.pem",
)

You can also disable certificate validation altogether, but beware that this is a security risk.

client = AuthenticatedClient(
    base_url="https://internal_api.example.com", 
    token="SuperSecretToken", 
    verify_ssl=False
)

Things to know:

1. Every path/method combo becomes a Python module with four functions:

   1. sync: Blocking request that returns parsed data (if successful) or None
   2. sync_detailed: Blocking request that always returns a Request, optionally with parsed set if the request was successful.
   3. asyncio: Like sync but async instead of blocking
   4. asyncio_detailed: Like sync_detailed but async instead of blocking

2. All path/query params, and bodies become method arguments.

3. If your endpoint had any tags on it, the first tag will be used as a module name for the function (my_tag above)

4. Any endpoint which did not have a tag will be in pykiwiai.api.default

Advanced customizations

There are more settings on the generated Client class which let you control more runtime behavior, check out the docstring on that class for more info. You can also customize the underlying httpx.Client or httpx.AsyncClient (depending on your use-case):

from pykiwiai import Client

def log_request(request):
    print(f"Request event hook: {request.method} {request.url} - Waiting for response")

def log_response(response):
    request = response.request
    print(f"Response event hook: {request.method} {request.url} - Status {response.status_code}")

client = Client(
    base_url="https://api.meetkiwi.ai",
    httpx_args={"event_hooks": {"request": [log_request], "response": [log_response]}},
)

# Or get the underlying httpx client to modify directly with client.get_httpx_client() or client.get_async_httpx_client()

You can even set the httpx client directly, but beware that this will override any existing settings (e.g., base_url):

import httpx
from pykiwiai import Client

client = Client(
    base_url="https://api.meetkiwi.ai",
)
# Note that base_url needs to be re-set, as would any shared cookies, headers, etc.
client.set_httpx_client(httpx.Client(base_url="https://api.meetkiwi.ai", proxies="http://localhost:8030"))

Building / publishing this package

This project uses Poetry to manage dependencies and packaging. Here are the basics:

1. Update the metadata in pyproject.toml (e.g. authors, version)
2. If you're using a private repository, configure it with Poetry
   1. poetry config repositories.<your-repository-name> <url-to-your-repository>
   2. poetry config http-basic.<your-repository-name> <username> <password>
3. Publish the client with poetry publish --build -r <your-repository-name> or, if for public PyPI, just poetry publish --build

If you want to install this client into another project without publishing it (e.g. for development) then:

1. If that project is using Poetry, you can simply do poetry add <path-to-this-client> from that project
2. If that project is not using Poetry:
   1. Build a wheel with poetry build -f wheel
   2. Install that wheel from the other project pip install <path-to-wheel>

Regenerating the client

The client is generated from the live API's OpenAPI document with openapi-python-client. _generated/ holds the raw generator output; pykiwiai/ is that same output with our patches applied on top.

# 1. Fetch the spec (the endpoint requires a bearer token)
curl -H "Authorization: Bearer $TOKEN" https://api.meetkiwi.ai/v1/openapi.json -o openapi.json

# 2. Force servers[0] to api.meetkiwi.ai
python3 scripts/patch_openapi.py openapi.json openapi.patched.json

# 3. Generate
uvx --from openapi-python-client openapi-python-client generate \
    --path openapi.patched.json --config openapi-python-client.yaml \
    --output-path _tmp --overwrite

# 4. Land the raw output, then apply our patches to the shipped package
rm -rf _generated pykiwiai && cp -R _tmp/pykiwiai _generated && cp -R _tmp/pykiwiai pykiwiai
python3 scripts/postprocess_client.py pykiwiai

# 5. Bump the version and build
poetry version <new-version>
poetry build

scripts/postprocess_client.py is idempotent and applies four patches:

• defaults base_url to https://api.meetkiwi.ai on both clients
• makes AuthenticatedClient.token keyword-only
• guards the typing.Self import so the package still imports on Python 3.10
• swaps datetime.fromisoformat for dateutil.isoparse, because 3.10's fromisoformat rejects the trailing Z the API returns on every timestamp

The generator reports pre-existing warnings for /v1/schedules/*, the CronSchedule / UrlResponse duplicate model names, and two /v1/vectorstore endpoints; those schemas come through partially or not at all.