seerlens 1.3.0

Send your Python app's LLM calls to Seerlens, the local DevTools for AI calls.

seerlens (Python)

Send your Python app's LLM calls to Seerlens. Traces go out as OpenTelemetry GenAI spans, so they land in the same dashboard as the .NET ones.

pip install seerlens

import seerlens

seerlens.configure("http://localhost:5005")

with seerlens.trace("answer ticket", model="gpt-4o") as span:
    reply = my_llm(prompt)
    span.complete(prompt=prompt, completion=reply, input_tokens=40, output_tokens=12)

seerlens.flush()  # before a short script exits

Or record a call you already made:

seerlens.record(model="gpt-4o", prompt="hi", completion="hello",
                input_tokens=10, output_tokens=5, duration_ms=820)

Traces are sent on a background thread. If the collector is down the trace is dropped; it never blocks or throws into your app.

No third-party dependencies. Run the example against a running collector:

python example.py