Integrations

TraceCraft integrates with popular LLM frameworks and cloud platforms.

No custom integration code required

Point any OTLP-instrumented app at TraceCraft and you’re done:

tracecraft serve --tui
OTEL_EXPORTER_OTLP_ENDPOINT=http://localhost:4318 python your_app.py

Works with OpenLLMetry, LangChain, LlamaIndex, DSPy, PydanticAI, and any standard OpenTelemetry SDK — no TraceCraft-specific code needed.

Framework Integrations

Automatic tracing for LangChain chains, agents, and tools

Seamless integration with LlamaIndex query engines and agents

Native support for PydanticAI agents and tools

Wrapper for Claude Agent SDK with automatic tracing

Automatic tracing for OpenAI and Anthropic SDKs

Receive traces from any OTLP source with simple setup

AWS AgentCore, Azure AI Foundry, GCP Vertex Agent Builder

The simplest path requires no TraceCraft-specific code — just set the OTLP endpoint:

OTEL_EXPORTER_OTLP_ENDPOINT=http://localhost:4318 python your_langchain_app.py

Or use auto-instrumentation with one line:

import tracecraft
tracecraft.init(auto_instrument=True)   # patches LangChain automatically

For richer span context using the native adapter, see LangChain Integration.

Set the OTLP endpoint if LlamaIndex is already emitting OTel traces:

OTEL_EXPORTER_OTLP_ENDPOINT=http://localhost:4318 python your_llamaindex_app.py

Or enable auto-instrumentation:

import tracecraft
tracecraft.init(auto_instrument=True)   # patches LlamaIndex automatically

For the native span handler, see LlamaIndex Integration.

Set the OTLP endpoint for any PydanticAI app already emitting OTel traces:

OTEL_EXPORTER_OTLP_ENDPOINT=http://localhost:4318 python your_pydanticai_app.py

Or use auto-instrumentation:

import tracecraft
tracecraft.init(auto_instrument=True)

For the native span processor, see PydanticAI Integration.

Set the OTLP endpoint for Claude SDK apps already emitting OTel traces:

OTEL_EXPORTER_OTLP_ENDPOINT=http://localhost:4318 python your_claude_app.py

Or use auto-instrumentation:

import tracecraft
tracecraft.init(auto_instrument=True)

For the full adapter with tool-level tracing, see Claude SDK Integration.

Framework	Use When	Installation
LangChain	Building with LCEL chains	`tracecraft[langchain]`
LlamaIndex	Using query engines or RAG	`tracecraft[llamaindex]`
PydanticAI	Type-safe agent development	`tracecraft[pydantic-ai]`
Claude SDK	Claude-powered agents	`tracecraft[claude-sdk]`
Auto	Using OpenAI/Anthropic directly	`tracecraft[auto]`
OTel Receiver	Receiving OTLP traces	`tracecraft[receiver]`
Custom	Rolling your own	Base package

Explore integration guides: