Automation (Jira & Studio) cannot target Sandbox environments critical gap for safe AI/Agent testing

[Kirk Acklin]

Hi Atlassian Community,

I'm running into what I believe is a significant gap in the automation platform, and I'd love to hear if others have experienced this and whether there's a known workaround or a fix on the roadmap.

The Situation

Our organization has a sandbox Atlassian site (yourorg-sandbox.atlassian.net) that we spun up specifically to safely test automations and Rovo AI agents before enabling them in production. This is exactly the responsible approach Atlassian recommends — test in sandbox, promote to production.

The Problem

When building an automation flow in Studio (or Jira Automation), the Jira project trigger dropdown only shows projects from the production site — even when Studio is accessed from the sandbox site URL. There is no way to scope a trigger to a sandbox Jira environment.

As a result:

When we enabled our flow for testing, it immediately fired on production tickets, not the sandbox tickets we created for testing.

The sandbox tickets were completely ignored.

We had no safe way to validate the Rovo Agent's behavior before it touched live customer data.

Why This Matters

This gap is especially critical now that Atlassian is pushing AI-powered automation (Rovo Agents in Studio flows). The stakes of an untested agent running on production tickets are much higher than a simple field-update automation. Teams need a safe, isolated environment to validate agent behavior before production rollout — and right now, that path doesn't exist cleanly.

What Would Help

Allow automation triggers (in both Jira Automation and Studio) to target projects from sandbox/non-production Atlassian sites within the same organization.

Or, provide a first-class "sandbox mode" for automation flows that prevents them from firing on production data while testing is in progress.

At minimum, document a supported testing path for Studio + Rovo Agent automations that doesn't require enabling flows in production.

Current Workarounds (None are ideal)

Using a label condition to scope which tickets the agent sees — functional but fragile and requires manual ticket tagging.

Keeping the flow disabled and testing manually — provides no real automation validation.

There is no way to point the trigger at the sandbox Jira project specifically.

Has anyone else hit this wall? Any workarounds that have worked well for your team? And Atlassian folks — is this on the roadmap?

Thanks in advance.