Well Snort has rules for it. So I can detect and block skype activity. Personally, lacking any directive from upon high, I've not blocked it. I only log installation and usage with snort.
Code:
alert tcp $HOME_NET any -> $EXTERNAL_NET $HTTP_PORTS (msg: "BLEEDING-EDGE Policy Skype VOIP Checking Version (Startup)"; flow: to_server,established; uricontent:"/ui/"; nocase; uricontent:"/getlatestversion?ver="; nocase; reference:url,www1.cs.columbia.edu/~library/TR-repository/reports/reports-2004/cucs-039-04.pdf; classtype: policy-violation; sid: 2001595; rev:6; )
alert tcp $HOME_NET any -> $EXTERNAL_NET $HTTP_PORTS (msg: "BLEEDING-EDGE Policy Skype VOIP Reporting Install"; flow: to_server,established; uricontent:"/ui/"; nocase; uricontent:"/installed"; nocase; reference:url,www1.cs.columbia.edu/~library/TR-repository/reports/reports-2004/cucs-039-04.pdf; classtype: policy-violation; sid: 2001596; rev:6; )
alert tcp $HOME_NET any -> $EXTERNAL_NET $HTTP_PORTS (msg: "BLEEDING-EDGE POLICY Skype User-Agent detected"; flow:to_server,established; pcre:"/User-Agent\:[^(\n|\r)]+Skype/i"; classtype: policy-violation; sid:2002157; rev:1;)
If your proxy allows blocking clients via User Agent that would be effective. Squid/Dansguardian certainly can.