Every day, I observe the IDE attempting to run search commands or node scripts through the console. The problem is that 99% of the time, the command or script doesn’t actually run and remains in an eternal “Running” state. Sometimes I have to press “cancel,” but the IDE’s behavior is different each time. It either tries to run the command again (sometimes it succeeds), or sometimes it completely aborts the task! I’ve also seen more than once that when running a command, the IDE spontaneously writes characters to the console, making the command invalid. For example, it often writes “ccnpm run build.” Generally, there are many problems with running commands: it runs from the wrong directory, appends characters, and does everything correctly, but the command doesn’t actually start running, and pressing “cancel” often aborts the task.
1 Like
Add to global Rules:
ENV: Windows. Use cmd /c for all shell execs to force process termination (EOF). Avoid interactive shells.
I fix this just by turn off and turn on the app again
IDE It almost always ignores the rules. Sometimes it ignores reading the rules file even if you specify it as mandatory via @Rules.
The situation is evolving. Previously, the agent’s commands would “freeze” selectively, but now my agent’s execution of any commands has completely stopped. I’ve created a new bug report in continuation of this thread: Agent terminal deadlock - stuck on "Running" for all commands
same issue for me on windows …
any updates from google team?