tidal-telemetry 0.1.0

One-line LLM observability for Tidal Telemetry - auto-instruments 50+ AI libraries and streams OpenTelemetry traces to your Tidal project.

tidal-telemetry

One-line LLM observability. Auto-instruments 50+ AI libraries (OpenAI, Anthropic, LangChain, LlamaIndex, Ollama, vector DBs, agent frameworks, ...) and streams OpenTelemetry traces - latency, tokens, cost, prompts - to your Tidal Telemetry project.

Install

pip install tidal-telemetry

Use

import tidal_telemetry as tidal

tidal.init()  # reads TIDAL_API_KEY + TIDAL_PROJECT from the environment

export TIDAL_API_KEY="<your project token>"
export TIDAL_PROJECT="<org-project>"
python your_app.py

Or configure in code:

tidal.init(api_key="tt_...", project="acme-production", application_name="checkout-bot")

Trace your own functions too:

@tidal.trace
def handle_request(prompt):
    ...

Your API key and project slug are on the project's onboarding page: https://openlit.llmcfo.com → Settings → Organisation → Projects.

Environment variables

Variable	Meaning
TIDAL_API_KEY	Project token (required unless passed to init)
TIDAL_PROJECT	Project slug, e.g. acme-production
TIDAL_ENDPOINT	Ingest base URL override (defaults to the hosted endpoint)
OTEL_SERVICE_NAME	Service name shown in dashboards

Built on the OpenLIT SDK (Apache-2.0); every openlit.init keyword argument passes through unchanged.