One line of code. Every framework.
Python and TypeScript SDKs that trace, evaluate, and optimize your agents in production. OpenTelemetry-native, framework-agnostic, and installable from PyPI and npm.
pip install langwatch
import langwatch
@langwatch.trace()
def run_agent(query: str) -> str:
# your agent code, untouched
...$ langwatch login · then run your agent. traces show up live in the platform.
Five minutes. From clone to first trace.
Install the SDK, set one environment variable, wrap your agent. Traces, evaluations, and prompt history land in the platform without changing your call sites.
- 01Install the SDKpip install langwatch · npm i langwatch
- 02Set your API keyexport LANGWATCH_API_KEY=… (or langwatch login)
- 03Wrap your agent@langwatch.trace() / lw.trace("triage-agent", async () => …)
- 04Add an eval, optionallylangwatch.evaluation.run("ragas/faithfulness", …)
Plays nicely with your stack.
Languages
Python SDKTypeScript SDK
Wire format
OpenTelemetry
Agent frameworks
OpenAI AgentsAnthropicCrewAILangGraphLangChainPydantic AIAgno
Model gateways
LiteLLMAWS BedrockAzure OpenAI
Optimization
DSPy
Other
REST APICLI