tvisha 0.1.0

Tvisha observability SDK — one-line tracing for LLM applications

tvisha

One-line observability for LLM applications. Traces LangChain, LangGraph, and any OpenTelemetry-instrumented code and sends data to the Tvisha platform.

Installation

pip install tvisha

# With automatic LangChain/LangGraph instrumentation:
pip install tvisha[langchain]

Quick start

import tvisha

tvisha.init(
    project_name="my-agent",
    api_key="tvisha_live_myproj_abc123...",
    otlp_endpoint="https://traces.tvisha.io",
)

# All LangChain / LangGraph calls are now traced automatically.

Configuration

Parameter	Env var fallback	Default
project_name	—	(required)
api_key	TVISHA_API_KEY	None (no auth header)
otlp_endpoint	TVISHA_ENDPOINT	http://localhost:8082

Arguments passed to init() take precedence over environment variables.

Environment-variable-only setup

export TVISHA_API_KEY="tvisha_live_myproj_abc123..."
export TVISHA_ENDPOINT="https://traces.tvisha.io"

import tvisha
tvisha.init(project_name="my-agent")