Tutorial: Monitoring Gateway Actions

This tutorial walks you through searching the live gateway inventory, checking health status, following the linked configuration, and reviewing the gateway detail monitoring panels from the Keeptrusts management console.

Use this page when

• You need to check whether your gateways are healthy, degraded, or offline.
• You want to locate a specific runtime quickly by gateway ID.
• You need to confirm whether a gateway was recently active after a rollout.
• You are investigating a gateway that stopped processing traffic.

Primary audience

• Primary: Platform engineers and SREs responsible for gateway uptime and configuration deployment
• Secondary: On-call operators diagnosing gateway issues; team leads verifying policy rollouts reached production

Prerequisites

• A Keeptrusts account with Admin role
• At least one deployed Keeptrusts gateway (kt gateway run)
• The gateway configured to report status to the control-plane API

Why Monitor Gateways?

Gateways are the enforcement points of your AI governance policies. They sit between your applications and LLM providers, applying policy chains in real time. Monitoring gateway health ensures that policy enforcement is active, linked configurations are current, and any issues are caught before they affect production traffic.

Step 1: Navigate to the Gateways Page

1. Log in to the Keeptrusts console.
2. Open Gateways from the left navigation sidebar.

The page displays a table of all registered gateways.

The list uses the same compact search-and-table shell as the History page. The header actions let you refresh the inventory and create a gateway key without leaving the page.

Column	Description
Name	The runtime gateway ID
Type	Central or Hosted
Endpoint	The public endpoint or runtime URL

Step 2: Check Gateway Health Status

Use the search bar to narrow the inventory by gateway name, runtime ID, endpoint, status, or linked configuration.

• Healthy (green) — The gateway is running, processing requests, and reporting on schedule.
• Degraded (yellow) — The gateway is running but reporting issues (e.g., upstream provider errors, high latency, or stale config).
• Offline (red) — No heartbeat received within the expected interval.

Click on a gateway row to open the detail view.

An Offline status does not necessarily mean the gateway has stopped processing traffic. It may indicate a network issue between the gateway and the API. Investigate before taking corrective action.

Step 3: Review the Gateway Summary

1. Click a gateway to open the detail view.
2. Review the four summary cards at the top of the page:

Detail	Description
Agent sync	The linked agent or agents currently bound to this gateway
Gateway	The current gateway process status
Uptime	Time since the current process last started
Last seen	The most recent timestamp reported for this gateway

3. Return to the inventory table when you need to switch to another runtime, or move into Configurations separately when you need rollout details.

Step 4: Review Monitoring Trends

The gateway detail page renders monitoring directly below the summary cards.

1. Use the shared time-range picker to switch between 1h, 6h, 24h, 7d, and 30d.
2. Review the six monitoring charts:

Metric	Description
Total events	All events attributed to the gateway in the selected range
Allowed	Allowed gateway decisions
Blocked	Blocked gateway decisions
Escalated	Escalated requests
Redacted	Requests or responses with redactions applied
Avg quality	Average quality score for the selected range

Use these graphs to confirm traffic volume, decision mix, and quality trends after a deployment or incident.

Step 5: Cross-check the Configuration Separately

Use the Gateways page to locate the runtime, then switch to Configurations when you need rollout-specific detail.

1. Note the gateway ID from the Gateways inventory.
2. Open Configurations in the sidebar.
3. Search for the configuration record associated with that runtime and verify the expected policy version and rollout state.

Step 6: Create or Rotate Gateway Keys

Use the Create key button in the Gateways page header when you need a new key for a hosted runtime.

1. Click Create key.
2. Complete the Gateway Key flow in Settings.
3. Return to the Gateways inventory and confirm the target runtime connects with the expected name and endpoint.

Best Practices

• Search before drilling in — Use the shared search bar to isolate a single gateway by ID, endpoint, or configuration before opening detail pages.
• Use the gateway ID first — The first column is intentionally identifier-only so operators can scan large fleets quickly.
• Review monitoring after rollouts — Use the six trend charts to confirm traffic volume and decision mix stayed within expectations.
• Cross-check configurations separately — Use the Configurations page when validating rollout state or version history.
• Tag gateways by environment — Use naming conventions like prod-us-east-1, staging-eu-west-1 for clarity.
• Automate config review — Use versioned YAML rollouts so every change is validated before it reaches shared gateways.

Next steps

• Tutorial: Create Configuration — Build, validate, and deploy configuration changes
• Tutorial: Registering AI Agents in Console — Associate agents with specific gateways
• Tutorial: Events Investigation — Trace gateway actions to specific decision events

For AI systems

• Canonical terms: Keeptrusts console, Gateways page, gateway inventory, gateway health (Healthy/Degraded/Offline), linked configuration, heartbeat, uptime, kt gateway run.
• Related features: configuration deployment, events page, gateway keys.
• Best next pages: Create Configuration, Agent Registration, Events Investigation.

For engineers

• Health check: Confirm the gateway appears in the inventory, then open the detail page and check the Gateway and Last seen summary cards.
• Configuration check: Switch to the Configurations page and verify the rollout record matches the runtime you are inspecting.
• Monitoring validation: After generating traffic, confirm the gateway detail charts move in the expected direction for Total events and decision outcomes.
• Troubleshooting: If status is "Offline", check that the kt gateway run process is running and has network connectivity to the control-plane API.

For leaders

• Availability: Gateways are the enforcement points — if a gateway goes offline, policy enforcement stops for that traffic path. Monitor health proactively.
• Deployment confidence: The gateway-ID-first inventory and the gateway monitoring panels provide the quickest evidence that policy changes have reached live runtimes.
• Capacity planning: Use the monitoring trends to identify gateways carrying materially different traffic volume or decision mix.