Step level context
Follow task status, trace identifiers, screenshots, and artifact manifests from the same run.
Trace keeps the record around a run: what the browser saw, what the agent did, and which artifacts were produced. It gives integration failures a shared surface for engineering, QA, and agents to inspect.
Follow task status, trace identifiers, screenshots, and artifact manifests from the same run.
Trace access is organization scoped, and artifact reads use authenticated or expiring references rather than an open bucket.
MCP tools expose trace listing, retrieval, bounded slices, diffs, and replay triggers for authorized clients.
See where a checkout changed from payment method to confirmation, where a verification email stopped arriving, or which callback assertion failed. The record makes a rerun more deliberate.
payment_intent.succeededOpen checkoutScreenshot captured200
Confirm paymentPOST /webhooks/payment200
Receive receiptTest inbox matchedready
Trace provides the structured record needed to investigate and reproduce supported failures. Replay availability depends on the workflow and deployment.
Talk through your use case4242 4242 4242 424208 / 28 · Test cardPOST /webhooks/payment signature verifiedHave a specific workflow in mind?
Let’s talk it through
Screenshots and trace artifacts are available in supported runs. Video and HLS replay embedding are still in beta; contact us about your workflow.
Yes, the hosted composite MCP endpoint exposes organization scoped Trace tools, subject to credentials and scopes.
Trace treats DOM, network, and console content as sensitive. Access is scoped and artifacts are served through authorization checks.
Choose a concrete workflow. We’ll help you get the right setup for your application.