Agent Cross-App Calling

Both Starcat and your project’s Starcat Developer can make cross-app calls for you. Instead of writing SDK code yourself, you can ask the agent to call another app’s API, discover what endpoints are available, or move data between apps. There are two situations where this happens, and they behave differently.

​

Starcat Operating Your Apps

When you chat with Starcat at the organization level — including the chat sidebar next to an app you’ve opened — it can operate your apps directly:

• Call an app’s API — “Get the latest orders from the POS app”
• Discover endpoints — “What can I call on the invoicing app?”
• List apps — “Which apps in our org can the agent operate?”

Starcat acts as itself, on your behalf — it isn’t standing in for one of your apps. For it to call an app, that app must turn on Allow agent access (see below). It does not use app-to-app grants.

​

Environment

When you’re viewing an app in the app shell, the agent operates the app in the environment you’re looking at — if you’re viewing production, it calls production; if you’re viewing a preview, it calls the preview. When the agent isn’t tied to an app you’re viewing, it defaults to production. You can override this by asking — “try this against the sandbox” — but you normally don’t need to think about it.

​

Starcat Developer Inside a Project

When you chat with the agent inside a specific project, it can call other apps too — for example to test an integration you’re building:

• Call another app’s API — “Call the inventory app’s /api/products endpoint”
• Choose the environment — “Call the CRM app’s sandbox to test the new endpoint”

Here the agent calls as the project you’re working in, using that project’s deployment credentials. It follows the same permission rules your app’s code would — it needs a grant (or allow-all access) to the target app, and it can only originate calls from the environments allowed by that project’s Developer-agent callable environments setting.

​

Allowing Starcat to Operate an App

By default, Starcat cannot operate an app. Open the app’s Settings → Cross-App tab and turn on Allow agent access. Once it’s on, choose which of the app’s environments the agent may call: Narrow this set to keep the agent away from environments you don’t want it touching. For example, allow Preview and Sandbox only while you’re still validating what the agent does.

​

Endpoint Discovery

Before making a call, the agent can discover what endpoints a target app exposes — useful when you don’t remember the exact path or want to see what an endpoint expects. Ask the agent things like:

• “What endpoints does the project-management app have?”
• “Show me the API routes on the invoicing app”
• “What does the /api/tasks endpoint expect as input?”

The agent reads the target app’s published endpoint list, including any documentation added alongside the API routes.

​

What Still Applies

Agent calls go through the same checks as SDK calls — the agent can’t bypass them:

• The endpoint must be published (or the app must have Allow all endpoints on)
• Built-in rate limits apply
• For Starcat, the target app must have Allow agent access on for the requested environment
• For the developer agent, the calling project needs a grant or allow-all access to the target

If a call is denied, the agent tells you why and what to enable to allow it.

​

Examples

Operate an app you’re viewing:

“Mark order #1043 as refunded.” (said to the Starcat sidebar while viewing the orders app)

Move data across apps:

“Get all customers from the CRM app and create corresponding contacts in the support app.”

Build and test an integration (from a project chat):

“Call the payments app’s /api/webhooks/test endpoint from this project’s sandbox and show me the response.”

Discover and explore:

“List the apps the agent can operate, then show me what endpoints the top-selling app has.”