Antigravity Built-in Command for sending a query via a chat to the model

Google Antigravity
1

We are authors of Spring Boot Tools extension that is installable into Antigravity. One of the features of the extension is CodeLens over some pieces of code that states “Explain with AI” which sends a query via the chat UI to the model.
We’d like to add support for this feature for Antigravity as well (already did Cursor). For this we’d like:

  1. String id of the command that sends a query via chat UI to the AI (“antigravity.sendTextToChat“ ?)
  2. What are the parameters for the command? If its an object expected the scheme of the object would be neccessarry.
2

Hello @Alex_Boyko, welcome to AI Forum!

Thank you for your interest in integrating the Spring Boot Tools extension with Antigravity. We would welcome a more detailed overview of the extension, as well as any documentation or project references you can provide. Once received, I will present this information to our engineering team for further evaluation.

4

Hey @Abhijit_Pramanik ,

thanks for coming back to us on this. Unfortunately, it looks like the answer that Alex posted a short while ago that included the links to the necessary information somehow got flagged as spam (or something similar), removed, and his account is blocked now - so that he can’t provide any new answers anymore. Can you have a look at unlock his account again?

In addition to that, you can find the “Spring Boot Tools” extension on the openvsx marketplace, which also includes links to the GitHub repo, etc. I won’t paste direct links here to avoid being banned, too. But I am sure you can find the extension easily there.

To describe the feature in more detail:
We analyze the source code in the editor and insert code lenses at specific locations to indicate that AI could help here with explaining or transforming something. When the user clicks on the code lens, we call the mentioned API to send a dynamically crafted prompt to the AI chat. It is basically a code-lens-based shortcut to avoid having the user copy-paste things into the AI chat.

We implemented this already for Copilot in VSCode, Copilot in Eclipse, and Cursor - and would love to support Antigravity here as well. Therefore the ask to provide this API.

If you tell me know to provide direct links without being banned from this forum, we would be happy to share more details anytime.

5

@Abhijit_Pramanik Any news on this one? Would love to continue the work on our side on this and add AntiGravity to the list of tools that the Spring Tools support, so would love to make some progress on this one here.

6

Hi @Martin_Lippert, welcome to AI Forum!
Sorry for the delay in responding. Thank you for sharing more details. I have forwarded your query on the Spring Boot Tools extension with the relevant internal team.
I’ll try to get Alex unblocked. His account was blocked due to posting links that were flagged as spam somehow as you guessed.

7

Hi @Abhijit_Pramanik , thanks a lot for coming back to me here, much appreciated. Would indeed be awesome if Alex account could be unblocked again.

And let me know what the next steps with regards to our question are. Happy to provide more details, feedback, suggestions, etc, anytime.

Cheers
Martin

8

Hey @Abhijit_Pramanik , any news on this? (both unblocking Alex account as well as feedback from the team around the command?)

9

I’d like to add a broader developer perspective to this discussion.

The current request around sending text to the chat highlights a deeper limitation: extensions don’t have a proper way to programmatically interact with the chat system as a first-class concept.

From our experience building extensions, the main gap is not just a single command, but the lack of a minimal API to control conversations. For example:

  • Starting a new chat session programmatically

  • Referencing or reusing an existing conversation (via an ID)

  • Forking a conversation into a new context

  • Sending structured prompts with attached context (code, selections, metadata)

Today, without these capabilities, developers are pushed toward workarounds (like indirect input/output handling or reverse engineering behavior), which are fragile and not sustainable.

What would help is a small but well-defined API surface around chat orchestration. Even something minimal like:

antigravity.chat.createSession()
antigravity.chat.sendMessage(sessionId, payload)
antigravity.chat.forkSession(sessionId)

would already unlock many extension scenarios.

This would allow developers to build richer integrations directly inside Antigravity instead of relying on manual user interactions or external tooling.

10

Hi @beqodia

For VS code I’ve something like this:

	private async sendQueryToCopilot(query: string): Promise<boolean> {
		try {
			await vscode.commands.executeCommand('workbench.action.chat.open', {
				query: query,
				isPartialQuery: true
			});
		} catch (error) {
			console.error('Error sending query to AI chat:', error);
			vscode.window.showInformationMessage("Error happened while sending to AI chat");
			return false;
		}

		return true;
	}

I just put the string inside the chatbox, giving the user the option of to send or not, can we’ve something like this Antigravity?