DeepAgents

For optional help from your coding agent, use Set it up for me. It preserves your model connection and verifies a real trace after setup

Send DeepAgents traces to LiteLLM Lens using the runnable examples in this repository.

Prerequisites

You need Lens installed alongside LiteLLM, a key with model access, and a configured model alias. The swarm example needs a model that supports tool calls. In Lens > Traces > Set up tracing, click Generate tracing key and copy the Traces endpoint under Connection details. Ask your administrator for these if you cannot create a tracing key.

Install uv. It uses the checked-in Python version and resolves each example’s dependencies from its uv workspace.

Configuration

For a fresh checkout:

git clone https://github.com/BerriAI/litellm-lens-example.git
cd litellm-lens-example/deepagents
cp .env.example .env

If you already cloned the repository, run the remaining commands from deepagents/. Copy .env.example to .env if it does not exist, then set:

Variable	Value
LITELLM_GATEWAY_URL	Your gateway’s base URL without a trailing slash or /v1, for example http://localhost:4002
LITELLM_API_KEY	Your LiteLLM model key
LENS_URL	Copy Traces endpoint from Lens tracing setup and remove the final /v1/traces. Keep /lens-ingest if present. For example, http://localhost:4318 or https://gateway.example/lens-ingest
LENS_TRACING_KEY	The dedicated tracing key from Lens tracing setup
LITELLM_MODEL	A model alias configured on your gateway

The checked-in values target a local development gateway. Replace them for your deployment. Keep the exporter settings from .env.example; the examples configure their trace exporters in code. They send traces to LENS_URL/v1/traces with the tracing key as a bearer token.

Leave MOCK_LITELLM_GATEWAY_URL unset unless you intend to send an additional trace copy to the local recorder.

Run an example

Simple agent

A single deep agent answers one question.

uv run --env-file .env --package lens-deepagents-simple simple/main.py

See simple/main.py for the implementation.

Agent swarm

A coordinator delegates to search_agent and writer_agent through the built-in task tool.

uv run --env-file .env --package lens-deepagents-swarm swarm/main.py

See swarm/main.py for the implementation.

Verify the trace

After the example prints its answer, open Lens > Traces on your gateway and select the new run. Look for the run associated with research_agent. Inspect the input, output, and model spans. For the swarm, inspect the specialist activity described above; its exact span layout depends on the framework.

How tracing works

OpenInference instruments the LangChain runs used by DeepAgents. The examples explicitly use the Responses API; subagent runs nest under the coordinator’s task tool calls.

Troubleshooting

If model calls fail, check the gateway URL, key, and model alias. If an answer appears but the trace is missing, check the terminal for exporter errors and confirm the Lens ingestion service is reachable with your tracing key. A model call succeeding does not confirm that its trace export succeeded.

Use a gateway model alias that supports the Responses API and tool calls.