weckr-sdk 0.1.0

AI cost and margin intelligence for SaaS founders

weckr

Token-budget enforcement and observability for LLM apps. One line to wrap any OpenAI or Anthropic client.

Install

pip install weckr

Quick start

import os
from openai import OpenAI
from weckr import Weckr

wk = Weckr(api_key=os.environ["WK_API_KEY"])
client = wk.wrap(OpenAI())

resp = client.chat.completions.create(
    model="gpt-4o-mini",
    messages=[{"role": "user", "content": "Hello!"}],
)
print(resp.choices[0].message.content)

Every call is logged to your Weckr dashboard with token counts, latency, and cost. If you exceed your daily token cap, the next call raises WeckrCapError instead of silently burning money.

How it works

wk.wrap(client) returns a proxy that:

1. Checks your remaining budget before each call (cached, ~5ms overhead).
2. Forwards the call to the underlying client unchanged.
3. Fires a non-blocking log to api.weckr.dev with usage data.

Works with sync and async clients. Streaming is passed through transparently — usage is logged when the stream closes.

Anthropic

from anthropic import Anthropic
from weckr import Weckr

wk = Weckr(api_key=os.environ["WK_API_KEY"])
client = wk.wrap(Anthropic())

msg = client.messages.create(
    model="claude-3-5-sonnet-latest",
    max_tokens=1024,
    messages=[{"role": "user", "content": "Hello!"}],
)

Direct chat

If you don't want to wrap a client, use wk.chat directly:

text = wk.chat(
    provider="openai",
    model="gpt-4o-mini",
    messages=[{"role": "user", "content": "Hi"}],
)

wk.chat raises WeckrCapError if you're over budget.

Configuration

Env var	Purpose
WK_API_KEY	Your Weckr API key (starts with wk_)
OPENAI_API_KEY	Forwarded to the OpenAI client
ANTHROPIC_API_KEY	Forwarded to the Anthropic client

Get your key at weckr.dev.

License

MIT