- A Veritrix account and project
- The Veritrix SDK snippet added to your agent code
- A validated connection
- Live traces streaming into the dashboard
Step 1: Register on Veritrix
Head to app.veritrix.xyz and create an account. You can sign up with Google, GitHub, or email. Once your account is ready, you’ll land on the Veritrix dashboard.
Step 2: Open the Get Started page
From the dashboard, click the Start button in the top navigation. This takes you to the/get-started onboarding page, where Veritrix helps you generate a ready-to-paste SDK snippet.

Step 3: Choose how you want to instrument your agent
On the Get Started page, pick the setup path that matches where your agent code lives:Fresh Code
Start from a blank template Veritrix generates for you. Great for exploring the product before wiring it into a real project.
Google Colab Notebook
Get a snippet pre-configured for a Colab environment. Ideal for prototypes, demos, and experiments.
Existing Codebase
Drop the snippet into an agent you already have running in your own repo or service.

Step 4: Copy the generated code snippet
Based on your selection, Veritrix generates a code snippet pre-filled with your project’s API key and endpoint. Copy it and paste it into your agent code — as close as possible to where your agent starts running, so it can capture every LLM call, tool use, and handoff.
Add to your agent entrypoint
Call
veritrix.init() before you import or instantiate your LLM clients and agent framework. That’s how Veritrix auto-instruments everything downstream.Step 5: Validate the connection
Back on the Get Started page, click Verify. Veritrix will wait for the first trace from your agent and confirm the SDK is connected correctly. Run your agent once (locally, in Colab, or in your service) after adding the snippet, then hit Verify.
Step 6: View live traces
Every time your agent runs, Veritrix captures the run via the ingest API and displays it in the Traces tab in real time. Open the tab to see:- The full trace waterfall for each run
- LLM calls, tool executions, and agent handoffs as spans
- Latency, token usage, and cost per span

What’s next
- Explore the Trace Waterfall to understand each span in a run
- Learn how Veritrix models multi-agent systems in Agents & Handoffs
- Set up Alerting so you’re notified before failures compound
- Dive into SDK Configuration to tune the SDK for staging vs. production