verifai-llm-sdk 0.2.2

VerifAI Python SDK for LLM tracing and evaluation

VerifAI SDK

VerifAI SDK provides automatic tracing and evaluation for your AI/LLM applications. With one line of code, capture all LLM calls and send them to the VerifAI platform for analysis.

Installation

pip install verifai-sdk

Quick Start

import verifai
from openai import OpenAI

# 1. Initialize (one line!)
verifai.auto_instrument(
    project_name="my_project",
    capture_content=True
)

# 2. Use your AI libraries as usual
client = OpenAI()
response = client.chat.completions.create(
    model="gpt-4",
    messages=[{"role": "user", "content": "Hello!"}]
)

# Traces are automatically sent to VerifAI! ✨

Configuration

Set your API key as an environment variable:

export VERIFAI_API_KEY="your_api_key_here"
export VERIFAI_API_URL="http://localhost:8000"  # Optional

Supported Frameworks

• ✅ OpenAI
• ✅ Anthropic
• ✅ LangChain
• ✅ LlamaIndex
• ✅ Custom agents (via @verifai.trace decorator)

Manual Tracing

For custom agent logic:

@verifai.trace
def my_agent(query: str):
    # Your agent logic here
    return result

License

Apache 2.0