Bug Report: Severe memory leak triggered by "TypeScript (Native Preview)" extension

[KostyaTretyak]

🔎 Search Terms

Note: this issue is updated version of #63849

🕗 Version & Regression Information

Extension identifier: typescriptteam.native-preview
Extension version: 0.20260708.2
VS Code Version: 1.134
Operating system Version: Ubuntu 22.04

The bug reproduces with and without the "TypeScript 7 Nightly" extension

Identifier: typescriptteam.vscode-typescript-nightly
Version: 0.20260826.1

I used this extension to have the latest version of TypeScript (with TypeScript: Select TypeScript Version -> Use TypeScript 7).

⏯ Playground Link

No response

💻 Code

Steps to reproduce

1. Create a new empty folder and create a single blank file named index.ts inside it. It is important to have many other git repositories nearby (e.g. ../*).
2. Disable the "TypeScript (Native Preview)" extension in VS Code.
3. Open the newly created folder in VS Code.
4. Open the index.ts file (this step is required to trigger the behavior).
5. Check the system monitor: memory consumption remains normal/baseline (~4.1 GB total system load).
6. Enable the "TypeScript (Native Preview)" extension.
7. Observe the system monitor: memory consumption starts growing rapidly.
8. Disable the extension and run the "Restart Extensions" command (do not close VS Code yet).
9. Observe that the memory leak continues unabated (climbing up to ~14.6 GB / 87.5% RAM usage).
10. Close VS Code completely: the memory leak stops.

Important clarification: When I add the default package.json file to the folder using the following command, there is no memory leak.

npm init -y

🙁 Actual behavior

Memory usage spikes dramatically and is never reclaimed until the entire VS Code application is shut down.

Screenshots

• Baseline / Normal state (~4.1 GB):

• Leaked state (~14.6 GB after enabling/disabling the extension):

🙂 Expected behavior

Enabling the extension on an empty file should not cause massive memory consumption. Disabling the extension or restarting extensions should terminate the underlying leaked processes and free up the allocated RAM.

Additional information about the issue

The CPU is also used quite actively. On the penultimate screen you can see the memory consumption curve.

The moment VS Code closes

You see a spike first - this is when VS Code is closed. And then the memory does not return to normal.