# /robots.txt file for http://www.waag.ch/

# have to allow Googlebot here since it will index 
# but not visit the page. I don't want it to be
# indexed!

User-agent: Googlebot
User-agent: Ask Jeeves/Teoma 
Disallow: /analog/
Disallow: /bots/
Disallow: /botsv/
Disallow: /cssfiles/
Disallow: /fpdb/
Disallow: /graphics/
Disallow: /htmlarea/
Disallow: /include/
Disallow: /intern/
Disallow: /scripts/
Disallow: /style/
Disallow: /webmin/
# Do not exclude legacy links kept for a search engine friendly transition
# Bot must walk the link in order to find out it's out-dated

User-agent: *
Disallow: /analog/
Disallow: /bots/
Disallow: /botsv/
Disallow: /cssfiles/
Disallow: /fpdb/
Disallow: /graphics/
Disallow: /galerie/
Disallow: /htmlarea/
Disallow: /include/
Disallow: /intern/
Disallow: /scripts/
Disallow: /style/
Disallow: /webmin/
# Exclude legacy links kept for a search engine friendly transition
Disallow: /aktuell/
Disallow: /cd/

# common excludes for bad bots
# They're banned via .htaccess, if they obey robots.txt they won't trigger an alarm

User-agent: Art-Online
Disallow: /

User-agent: CherryPicker
Disallow: /

User-agent: Crescent Internet ToolPak
Disallow: /

User-agent: EmailCollector
Disallow: /

User-agent: EmailSiphon
Disallow: /

User-agent: EmailWolf
Disallow: /

User-agent: ExtractorPro
Disallow: /

User-agent: HTTPClient
Disallow: /

User-agent: Indy Library
Disallow: /

User-agent: Java
Disallow: /

User-agent: Microsoft URL Control
Disallow: /

User-agent: NexaBot
Disallow: /

User-agent: Nutch
Disallow: /

User-agent: VoilaBot
Disallow: /

User-agent: wavepluz
Disallow: /