Sign inSign up

iamkalio/traceflow-dashboard

By iamkalio

•Updated 7 months ago

Web UI for viewing traces from the [traceflow-ai](https://pypi.org/project/traceflow-ai/) SDK.

Image
Machine learning & AI
Monitoring & observability
0

1.4K

iamkalio/traceflow-dashboard repository overview

Traceflow is a local AI observability stack for LLM apps. The image runs the Traceflow dashboard (API + UI): it receives traces from the Python SDK and shows them in a simple UI (traces, spans, cost, latency). Backend is FastAPI with SQLite; no external database. Use it to inspect and debug LLM calls from your app without a cloud service.

How to run

docker run -p 8000:8000 iamkalio/traceflow-dashboard

Open http://localhost:8000 for the dashboard. In your app, install the SDK (pip install traceflow-ai[openai]), call traceflow_ai.init(endpoint="http://localhost:8000"), and use the OpenAI client as usual; traces will show up in the dashboard.

Tag summary

Content type

Image

Digest

sha256:32f3a6ad0…

Size

55.8 MB

Last updated

7 months ago

docker pull iamkalio/traceflow-dashboard