The 'logs' CLI command currently shells out to 'tail -f', which doesn't exist on Windows. Anyone trying to run the agent natively on Windows hits an error immediately when they try to follow the log output.
The fix is to replace the subprocess tail call with a pure Python implementation — open the file, seek to the end, and poll for new lines in a loop. The watchdog or similar library would also work, but a simple polling loop is probably fine here since this is a debug tool.
Relevant code is in backend/src/analytics_agent/cli.py in the logs command.
The 'logs' CLI command currently shells out to 'tail -f', which doesn't exist on Windows. Anyone trying to run the agent natively on Windows hits an error immediately when they try to follow the log output.
The fix is to replace the subprocess tail call with a pure Python implementation — open the file, seek to the end, and poll for new lines in a loop. The watchdog or similar library would also work, but a simple polling loop is probably fine here since this is a debug tool.
Relevant code is in backend/src/analytics_agent/cli.py in the logs command.