Router vulnerability note - MikroTik
MikroTik WebFig: a session token that survives the session
What happened
VulnCheck's advisory describes the mechanism precisely: makeSessionKey in
libucrypto.so derives WebFig session tokens from fixed salts, without channel- or session-specific
input, and the user service validates a token against the login store without checking the channel context. So a
token captured from one session keeps working in another, newly negotiated ECDH session — and it keeps working
until somebody clears the token store.
The advisory explicitly extends CVE-2025-61481, the cleartext-HTTP finding, and that is the honest way to read both. The transport problem was the door; this is the fact that a key left in the door stays valid. For an administrator who logs into WebFig from shared machines, the practical mitigation is the same as it was last year: HTTPS with a real certificate, HTTP off, and end the session rather than closing the tab.
Who is affected
RouterOS through 7.24.5 with the WebFig interface in use.
Check it yourself
Session behaviour and where tokens live
[admin@router] > /user active print # NAME GROUP ADDRESS LOGIN 0 admin full 10.44.9.2 web [admin@router] > /ip service print where name~"www" 7 www 80 0.0.0.0/0 8 www-ssl 443 0.0.0.0/0 letsencrypt-e1 [admin@router] > /user active remove [find where name!="yuri"]
Do this
- Wait for the fix, or use HTTPS and end sessions explicitly; a tab closed is not a logout on WebFig.
- Kill stale sessions after a shared-machine login:
/user active printand remove the ones that are not yours. - Disable the plain HTTP service now, since that is the exposure that makes a stolen token useful.
- Do not leave admin sessions open on a monitoring display that anyone can walk up to.
How we handle it
The two entries together — transport and token lifetime — are the strongest argument we have for putting a certificate on the box. We do it during the tunnel setup, and it costs nothing extra.