Skip to main content
Send traces to AgentMark from any OpenTelemetry-instrumented app, whether through a framework instrumentor or a raw OpenTelemetry setup. Point its exporter at AgentMark’s OTLP endpoint; there’s nothing to install on the AgentMark side.

​
Tracing

OpenTelemetry

The OTLP endpoint, authentication, and supported conventions

OpenInference

LangChain, LlamaIndex, CrewAI, DSPy, and more

OpenLLMetry

Traceloop and OpenLIT instrumentors

​
Frameworks

Per-framework setup for the most common agent and orchestration libraries. Each points an instrumentor at the same OTLP endpoint.

LangChain

LangChain and LangGraph: Python and JavaScript

CrewAI

CrewAI agents and crews: Python

LlamaIndex

LlamaIndex queries and agents: Python

DSPy

DSPy modules and optimizers: Python

OpenAI Agents SDK

OpenAI Agents SDK: Python and JavaScript

Vercel AI SDK

Vercel AI SDK v6: TypeScript

Pydantic AI

Pydantic AI agents: Python

AutoGen

Microsoft AutoGen AgentChat: Python

Agno

Agno agents: Python

Semantic Kernel

Microsoft Semantic Kernel: Python

smolagents

Hugging Face smolagents: Python

Haystack

Haystack pipelines: Python

Google ADK

Google Agent Development Kit: Python

Mastra

Mastra agents: TypeScript

​
Model providers

Trace a provider SDK directly. Each uses the matching OpenInference instrumentor pointed at the same OTLP endpoint.

OpenAI

OpenAI and Azure OpenAI: Python

Anthropic

Anthropic Claude: Python

Amazon Bedrock

Amazon Bedrock (boto3): Python

Google Gemini

Google Gemini: Python

Vertex AI

Google Vertex AI: Python

Mistral

Mistral: Python

Groq

Groq: Python

OpenAI-compatible

OpenRouter, Together, Fireworks, DeepSeek, Ollama, and more

​
Gateways

Trace calls routed through an LLM gateway. Each points an instrumentor at the same OTLP endpoint.

LiteLLM

LiteLLM: Python

Portkey

Portkey gateway: Python

​
Have questions?

Reach out any time: