DOCUMENTATION
AI NetShell documentation
Set up the terminal, connect a device, investigate a problem, and run an approved change.
Three questions, answered before you start
Read commands may run without approval. Configuration commands stay paused until the signed-in user approves the exact command set.
Can it break something?
Yes. A change you approve can still fail or interrupt traffic. No configuration command runs until the signed-in user approves the exact command set.
Can it see everything?
No. It sees only output from devices and sessions you connect. Recognized secret formats are redacted before a model request leaves the workstation, but unrecognized secrets may remain.
What happens if it is wrong?
If the proposed commands are wrong, decline them. After an approved change, AI NetShell reads the device again and offers rollback where the vendor recovery method supports it. Rollback can fail, so keep an independent path in.