Installation

Python 3.10 or later for the Python SDK, or Node.js 18 or later for the TypeScript SDK
A PandaProbe account with an API key

Some optional provider and framework packages require a newer Node.js version. Follow the engine requirements of the integrations you install.

pip install pandaprobe

uv add pandaprobe

npm install pandaprobe

yarn add pandaprobe

pnpm add pandaprobe

PandaProbe keeps provider and framework SDKs optional. Install only the packages used by your application.

Integration	Install command
OpenAI	`pip install "pandaprobe[openai]"`
Anthropic	`pip install "pandaprobe[anthropic]"`
Gemini	`pip install "pandaprobe[gemini]"`
Mistral	`pip install "pandaprobe[mistral]"`
LiteLLM	`pip install "pandaprobe[litellm]"`
AWS Bedrock	`pip install "pandaprobe[bedrock]"`
LangGraph	`pip install "pandaprobe[langgraph]"`
LangChain	`pip install "pandaprobe[langchain]"`
DeepAgents	`pip install "pandaprobe[deepagents]"`
Google ADK	`pip install "pandaprobe[google-adk]"`
Claude Agent SDK	`pip install "pandaprobe[claude-agent-sdk]"`
CrewAI	`pip install "pandaprobe[crewai]"`
OpenAI Agents SDK	`pip install "pandaprobe[openai-agents]"`

With uv, replace pip install with uv add. You can combine extras, for example uv add "pandaprobe[openai,anthropic,gemini]".

Integration	Additional packages
OpenAI	`openai`
Anthropic	`@anthropic-ai/sdk`
Gemini	`@google/genai`
Mistral	`@mistralai/mistralai`
AWS Bedrock	`@aws-sdk/client-bedrock-runtime`
LangGraph	`@langchain/langgraph @langchain/openai @langchain/core`
LangChain	`langchain @langchain/openai @langchain/core`
DeepAgents	`deepagents langchain @langchain/core @langchain/openai @langchain/langgraph @langchain/langgraph-checkpoint @langchain/langgraph-sdk` and a compatible `langsmith` release
Claude Agent SDK	`@anthropic-ai/claude-agent-sdk`
OpenAI Agents SDK	`@openai/agents`
Vercel AI SDK	`ai @ai-sdk/openai`

npm install pandaprobe openai
# or: yarn add pandaprobe openai
# or: pnpm add pandaprobe openai

LiteLLM, CrewAI, and Google ADK integrations are currently Python-only. Vercel AI SDK integration is TypeScript-only.

Minimum required setup for both SDKs:

export PANDAPROBE_API_KEY="your-api-key"
export PANDAPROBE_PROJECT_NAME="my-project"

You can also configure the SDK programmatically with Python’s pandaprobe.init(...) or TypeScript’s init({...}). See Project Configuration.

For the full list of variables, see Environment Variables.

Python

import pandaprobe
print(pandaprobe.__version__)

TypeScript

import { VERSION } from "pandaprobe";
console.log(VERSION);