Quickstart

Account to a populated dashboard in about five minutes.

Create your account

On first sign-in a tenant is created for you. A tenant is one isolated workspace with its own database. Your agents, costs and traces live in it, and nobody outside it can read them. You can create more later and invite your team; see Account & team.

Create an API key

Go to Account → API keys and click Create key.

The key is shown once, at creation. Only a hash is stored, so it cannot be displayed again. Copy it now. If you lose it, revoke it and create another.

Install the OTLP exporter

Once per Python environment. Your agent code is not modified.

Terminal

pip install opentelemetry-exporter-otlp-proto-http

The package is not optionalopentelemetry-exporter-otlp-proto-http is not bundled with google-adk. Without it the exporter has nowhere to send to and fails silently, which looks exactly like “TraptureIQ isn't working”.

Set the environment variables with export

Paste this into the terminal you will run the agent from, with your key in place of tiq_your_key_here. Change my-agent to your agent's name.

Local (adk web)

export OTEL_EXPORTER_OTLP_TRACES_ENDPOINT="https://api.traptureiq.com/v1/traces"
export OTEL_EXPORTER_OTLP_LOGS_ENDPOINT="https://api.traptureiq.com/v1/logs"
export OTEL_EXPORTER_OTLP_HEADERS="X-API-Key=tiq_your_key_here"
export OTEL_SERVICE_NAME="my-agent"
export OTEL_RESOURCE_ATTRIBUTES="deployment.environment=local"

Use export, not a .env file

adk web reads these variables once, at startup, from the terminal. Your agent's .env is loaded later, so values there are never seen and nothing is exported. Exports last for that terminal session; in a new terminal, paste them again.

Run your agent with adk web

In the same terminal, from the folder that contains your agent folder:

Terminal

adk web .

Tested on every ADK 2.x release and on 1.39.1; check yours with . Older 1.x releases send cost data but not the agent structure. Use adk web or adk api_server, not adk run, which exports nothing.

Ask it something, then check the dashboard

Allow 10–30 seconds for spans to arrive, then:

Observability → Traces shows the run with its tokens and cost; open it for the span timeline.
Observability → Cost starts drawing the spend chart.

The very first run on a new account sets up your workspace and can miss that window. If nothing appears, send a second message.

Deploying to Cloud Run, Agent Engine or GKE instead? Each takes the same variables a different way; see Connect your agent.

Then what?

Set a budget so spend cannot surprise you.
Run an eval against the agent, including the seeded security red-team sets.
Move a prompt into the prompt library so you can change it without redeploying.

Nothing showing up? Work through Troubleshooting, it is nearly always one of three things.