SOLVED - Antigravity hangs due to language_server_windows_x64 high memory consumption

I can maybe push a couple of prompts before the app hangs. I have noticed in these cases the language server consumes most of the available RAM. Any help appreciated.

EDIT
Ok I think the project just got too big and I was working in the root folder. I am now working in a sub folder and have no issue at the moment

4 Likes

This actually only served as a temporary fix, even the app in the subfolder got too big. I eventually identified the main cause to be the application source code files being too big. Some were over 1000 lines long. I instructed the agent to modularise the code as much as possible to prevent the language server having to index massive files and this appears to be a more permanent fix to my RAM issues.

Same problem and it started when I updated antigravity. Device is super hot and feels like it will melt if I work another 30 mins !!

Why on earth is this marked Solved?? It is not solved!!

3 Likes

it seems to be a memory leak + big issue of optimisation in working Agent (it’s my own case => CPU is over with Agent, draining my battery and launching refreshing ventilators everytime i use Antigravity Agent (even if a simple request like)

this issue is not solved, i agree with you

5 Likes

Hi,
I’m experiencing the same problem. My project isn’t that big, but the issue seems to have started after installing Docker/Phi-3 and Ollama. The following error keeps recurring in the Output Terminal window: Chrome Dev Tools constantly tries to establish an MCP server connection. I still haven’t been able to solve the problem. Any ideas?

(The following sentence is a separate, unrelated thought and not part of the problem description.) Output In Terminal “2026-02-11 10:54:16.623 [error] [Window] [Extension Host] [ExtensionServer] Failed to get Chrome DevTools MCP URL: Error: command ‘antigravity.getChromeDevtoolsMcpUrl’ not found
at r9i.n (vscode-file://vscode-app/c:/Users/XXXX/AppData/Local/Programs/Antigravity/resources/app/out/vs/workbench/workbench.desktop.main.js:XXX:XXXX)
at r9i.executeCommand (vscode-file://vscode-app/c:/Users/XXXX/AppData/Local/Programs/Antigravity/resources/app/out/vs/workbench/workbench.desktop.main.js:XXXX:XXXX)”

1 Like

I am having same problem. This exe is grabbing 1.5+ GB of Memory.

I wrote a hotfix extension to patch the error, but the IDE still crashes after a while :frowning: same issue

I wished it was solved, it’s eating memory like there’s no tomorrow

1 Like

Hi @Luigi_Simoncini ,
Thank you for reporting the issue.
Working from a subfolder is the recommended approach for large repos and usually resolves the issue,.We suggest to exclude large folders , restarting the Language server and limiting workspace scope to only what you’re actively editing.

I have the same problem. Big project, facing memory problems.
what does “subfolder” actually mean? new conversation?

haha. But that’s the only downside.

the actual solution to this is to have the builds/previews w.e. your output is with the 10k tiny files. put those outside of your workspace because antigravity is putting them in memory. I was get 18gb constantly till i moved it

1 Like

I faced that issue recently. The issue always happens when I edit the Markdown file in my workspace. The workspace is not too big, with only four applications; each application contains around 1,000 LOC only.

I want to share my experience — maybe it will help someone avoid the same situation.

I also use Antigravity, and I was running several projects in parallel (usually two or three at the same time). Everything worked fine and there were no issues with performance or memory.

At one point I asked my Antigravity agent to look at the root folder and clean things up a bit. It said that in order to do that it would need to reorganize a few folders, rename some things, and possibly merge a few skills. Overall it sounded like a normal structure optimization — nothing critical.

After these changes, I started experiencing serious memory problems. When I explained the situation to the agent and asked it to restore everything back to the original state, it was no longer able to do that.

Right now the situation is the following:
even when performing simple operations my RAM usage goes up to about 90%, even though I have 16 GB of RAM. Antigravity is currently running just in the browser, and memory usage is already around 80%.

At the same time I have another window open, but it’s just a simple conversation — I’m not reading large files, not working with code, and not running anything heavy. It’s basically just discussion.

So unfortunately I can confirm that the issue is still not resolved.

Hi there,

Just wanted to share my experience. I edited my Gemini.md file along with allowing run all commands in settings and after that noticed that a python subprocess spiked its RAM usage to massive levels. I killed the subprocess, removed the contents of Gemini.md yet it still spiked again (possibly because I didn’t restart the app but Gemini.md seems to be dynamically reloaded). So I put back the contents in Gemini.md, and checked settings, and the always allow/never allow commands settings had all been reset! The system seems to have returned to normal again (no spikes noticed) though. So I don’t have a definitive reason why it was spiking earlier but it seems related to the always allow/never allow commands settings. Oh, I had also changed a setting to enable explain and fix but that also had been reset. I will update if I discover anything else. Thanks!

Update: Still spiking. I have now removed Gemini.md contents. Confirmed all settings had been reset. Look at the insane spike!

Update 2: still happening after removing Gemini.md contents also. I bet if I don’t kill this python subprocess spawned by pytest.exe, it will keep growing. Slower spike now though.

Here is the pytest command that antigravity ran: pytest --tb=short *>&1 | Out-File -FilePath pytest_full_final.txt -Encoding utf8. It doesn’t spike right away when it is run. It takes a few minutes, then spikes rapidly.

Figured I could also drop my env details here if that provides a clue. But this has been sudden and it was working fine on this setup just a few hours ago (before I made Gemini.md and settings changes).

Antigravity Version: 1.20.6
VSCode OSS Version: 1.107.0 (user setup)
Commit: 135ccf460c67c4b900dc10
Date: 2026-03-16T23:21:29.372Z
Electron: 39.2.3
Chromium: 142.0.7444.175
Node.js: 22.21.1
V8: 14.2.231.21-electron.0
OS: Windows_NT x64 10.0.26200
Language Server CL: 884668262