Skip to main content
The SDK is pre-1.0 and its surface is still settling. The snippets below show the shape of the integration; check the SDK release notes for exact signatures before you pin a version.

​
1. Install

pip install visceral-ai
The distribution is visceral-ai; the import stays visceral.

​
2. Configure

Visceral reads its credentials from the environment. Your workspace API key comes from the dashboard (or the CLI). 
export VISCERAL_API_KEY="vsk_..."
By default the SDK reports to https://api.visceralai.dev.

​
3. Wrap your client

Wrap the LLM client your agent already uses. Visceral observes the traffic and reports metadata only — it never changes the response your agent receives. 
import visceral
from openai import OpenAI

client = visceral.wrap(OpenAI())

# use the client exactly as before
resp = client.chat.completions.create(
    model="gpt-4o",
    messages=[{"role": "user", "content": "hello"}],
)
That’s it — your agent behaves identically, and its LLM traffic now shows up in Visceral. If the backend is unreachable, the wrapped client falls straight through to the underlying provider (fail open).

​
Next steps

Instrumenting frameworks

LangChain, LangGraph, and other framework integrations.

How it works

What happens to the traffic once Visceral sees it.