Agents: cloud and private

Synthetic agents: cloud and private

Tests execute on agents. Cloud agents run from VerOps-managed regions for public targets. Private agents are a container you run inside your network to test internal apps, staging, and anything behind a firewall — from your users' vantage point.

Cloud agents

Nothing to install. When creating a test, pick one or more regions. Use multiple regions to detect geographic or CDN issues.

Installing a private (on-prem) agent

In Synthetics → Agents → Add private agent, generate a connection key.
Run the agent container where it can reach your target (the agent dials out to VerOps — no inbound ports needed):

docker run -d --name verops-synthetics-agent \
  --restart unless-stopped --shm-size=1g \
  -e VEROPS_AGENT_CONNECTION_KEY=<your-connection-key> \
  -e VEROPS_PLATFORM_URL=https://app.verops.io \
  verops/synthetics-agent:latest

The agent appears as Online within a few seconds.
In a test, set Run from to your private agent.

What can run where is decided by capabilities, not by cloud vs private. Every agent advertises what it supports (browser steps, screenshots, video, HAR capture, API checks), and the platform only dispatches a test to agents that advertise everything it needs. A private agent runs full browser tests — including screenshots and video — just like a cloud one. Cloud agents are shared, VerOps-managed locations and appear read-only in Agents; private agents are yours to manage, and their region/environment labels come from the settings you give them.

Sizing and reliability

Browser tests need more memory than API tests — give the container at least 1–2 GB and --shm-size=1g for Chromium.
Run two agents for redundancy if a journey is critical.
Agents auto-reconnect; monitor their Online status (you can even alert on it).

Troubleshooting

Agent offline: check outbound HTTPS to app.verops.io and that the connection key is correct.
Target unreachable: confirm the agent's host can resolve and reach the target (DNS, routes, firewall).