Merge pull request 'Improve SpamAssassin config, add custom rules' (#668) from email/spamassassin_rules into master

Reviewed-on: #668
Reviewed-by: Greg <greg@kosmos.org>

site-cookbooks/kosmos_blossom/templates/default/nginx_conf_blossom.erb

@@ -19,7 +19,7 @@ server {
   ssl_certificate <%= @ssl_cert %>;
   ssl_certificate_key <%= @ssl_key %>;
 
-  location ~ ^/(?<sha256>[a-f0-9]{64})(\.[a-zA-Z0-9]+)?$ {
+  location ~ "^/(?<sha256>[a-f0-9]{64})(\.[a-zA-Z0-9]+)?$" {
     proxy_set_header Host $host;
     proxy_set_header X-Real-IP $remote_addr;
     proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;

site-cookbooks/kosmos_email/recipes/spamassassin.rb

@@ -29,6 +29,12 @@ template "/etc/spamassassin/local.cf" do
   notifies :restart, "service[spamassassin]", :delayed
 end
 
+template "/etc/spamassassin/kosmos.cf" do
+  source "spamassassin_kosmos.cf.erb"
+  mode   0644
+  notifies :restart, "service[spamassassin]", :delayed
+end
+
 service "spamassassin" do
   action [:enable, :start]
 end

site-cookbooks/kosmos_email/templates/spamassassin_default.erb

@@ -30,4 +30,4 @@ PIDFILE="/var/run/spamd.pid"
 # Cronjob
 # Set to anything but 0 to enable the cron job to automatically update
 # spamassassin's rules on a nightly basis
-CRON=0
+CRON=1

site-cookbooks/kosmos_email/templates/spamassassin_kosmos.cf.erb

@@ -0,0 +1,64 @@
+###########################################################################
+# Kosmos custom SpamAssassin rules
+#
+# The rules below are designed to be individually weak but combine via meta
+# rules into strong, low-FP signals.
+###########################################################################
+
+# --- Individual signals --------------------------------------------------
+
+# URLs of the form https://////////////... (3+ slashes after the scheme).
+# Legitimate mailers never produce this; it is an obfuscation artefact.
+rawbody KOSMOS_MULTI_SLASH_URL /https?:\/{3,}/
+describe KOSMOS_MULTI_SLASH_URL URL with three or more consecutive slashes
+
+# Authoring-tool fingerprint left in the HTML by the spam toolchain.
+body KOSMOS_MSHTML_11_9600 /MSHTML 11\.00\.9600\.17037/
+describe KOSMOS_MSHTML_11_9600 HTML generated by MSHTML 11.00.9600.17037
+
+# Display name pattern used by the campaign.
+header KOSMOS_FROM_LUXURY_GOODS From:name =~ /Luxury (Watches|Bags|Handbags|Timepieces)\b/i
+describe KOSMOS_FROM_LUXURY_GOODS From display name advertises luxury goods
+
+# Base64-encoded unsubscribe links: return.php?p=<long base64>
+uri KOSMOS_RETURN_PHP_B64 /return\.php\?p=[A-Za-z0-9+\/=%]{20,}/
+describe KOSMOS_RETURN_PHP_B64 Base64-encoded return.php unsubscribe link
+
+# Fabricated "security" headers injected to evade heuristic filters.
+# No legitimate MTA or mailing-list manager emits these.
+header KOSMOS_FAKE_HDR_PHISHSIM    exists:X-PhishSimulator-Mode
+header KOSMOS_FAKE_HDR_DECEPTION   exists:X-Deception-Asset-Type
+header KOSMOS_FAKE_HDR_OBFUSCATION exists:X-Obfuscation-Trace-ID
+header KOSMOS_FAKE_HDR_QUARANTINE  exists:X-Quarantine-Reason-Code
+header KOSMOS_FAKE_HDR_TRUST       exists:X-Behavioral-Trust-Index
+
+# --- Scores for individual signals --------------------------------------
+score KOSMOS_MULTI_SLASH_URL   2.0
+score KOSMOS_MSHTML_11_9600    1.2
+score KOSMOS_FROM_LUXURY_GOODS 0.5
+score KOSMOS_RETURN_PHP_B64    1.5
+score KOSMOS_FAKE_HDR_PHISHSIM    1.0
+score KOSMOS_FAKE_HDR_DECEPTION   1.0
+score KOSMOS_FAKE_HDR_OBFUSCATION 1.0
+score KOSMOS_FAKE_HDR_QUARANTINE  1.0
+score KOSMOS_FAKE_HDR_TRUST       1.0
+
+# --- Meta rules ----------------------------------------------------------
+
+# Core campaign signature: luxury-goods From name + MSHTML fingerprint +
+# HTML-only body. Covers the bulk of the campaign corpus.
+meta KOSMOS_LUXURY_SPAM_CAMPAIGN (KOSMOS_FROM_LUXURY_GOODS && KOSMOS_MSHTML_11_9600 && MIME_HTML_ONLY)
+describe KOSMOS_LUXURY_SPAM_CAMPAIGN Luxury-goods From + MSHTML 11.00.9600 + HTML-only
+score KOSMOS_LUXURY_SPAM_CAMPAIGN 3.5
+
+# Luxury-goods From name combined with a suspicious URI signal or a
+# Spamhaus-listed relay.
+meta KOSMOS_LUXURY_SPAM_URI (KOSMOS_FROM_LUXURY_GOODS && (KOSMOS_MULTI_SLASH_URL || KOSMOS_RETURN_PHP_B64 || RCVD_IN_SBL_CSS))
+describe KOSMOS_LUXURY_SPAM_URI Luxury-goods From + suspicious URI or SBL relay
+score KOSMOS_LUXURY_SPAM_URI 2.5
+
+# Two or more fabricated "security" headers. Genuine mail never carries
+# these; their presence indicates a header-injection evasion kit.
+meta KOSMOS_FAKE_SECURITY_HEADERS (KOSMOS_FAKE_HDR_PHISHSIM + KOSMOS_FAKE_HDR_DECEPTION + KOSMOS_FAKE_HDR_OBFUSCATION + KOSMOS_FAKE_HDR_QUARANTINE + KOSMOS_FAKE_HDR_TRUST >= 2)
+describe KOSMOS_FAKE_SECURITY_HEADERS Two or more fabricated security headers
+score KOSMOS_FAKE_SECURITY_HEADERS 4.0

site-cookbooks/kosmos_email/templates/spamassassin_local.cf.erb

@@ -16,6 +16,37 @@ whitelist_auth <%= @whitelist_auth %>
 #    _CONTACTADDRESS_ in the report template)
 report_contact <%= @report_contact %>
 
+###########################################################################
+# Kosmos custom score overrides
+#
+# The Validity (Return Path / SenderScore) "certified sender" whitelists
+# (RCVD_IN_VALIDITY_CERTIFIED, RCVD_IN_VALIDITY_SAFE) hand out up to -5.0
+# of credit to sending IPs. These lists are commercially gamed and
+# routinely award -5.0 to IPs that are simultaneously listed on Spamhaus
+# SBL-CSS, SpamCop, MSPIKE and Validity's own RPBL. Neutralise them.
+###########################################################################
+score RCVD_IN_VALIDITY_CERTIFIED 0
+score RCVD_IN_VALIDITY_SAFE 0
+
+###########################################################################
+# Bayes hardening
+###########################################################################
+use_bayes 1
+bayes_auto_learn 1
+
+# Do not let Bayes learn from SpamAssassin's own result headers or from
+# Authentication-Results, which leak signal about prior scoring runs.
+bayes_ignore_header X-Spam-Flag
+bayes_ignore_header X-Spam-Status
+bayes_ignore_header X-Spam-Level
+bayes_ignore_header X-Spam-Checker-Version
+bayes_ignore_header Authentication-Results
+
+# Only learn ham when the message is clearly clean, and only learn spam
+# when it is clearly spam. The defaults (0.1 / 6.0) let marginally-spam
+# or marginally-ham messages poison the database.
+bayes_auto_learn_threshold_nonspam -1.0
+bayes_auto_learn_threshold_spam 8.0
 
 #   Add *****SPAM***** to the Subject header of spam e-mails
 #

site-cookbooks/kosmos_openresty/recipes/default.rb

@@ -3,7 +3,7 @@
 # Recipe:: default
 #
 
-node.normal['openresty']['log_formats']['json'] = '{"ip":"$remote_addr","time":"$time_local","host":"$host","method":"$request_method","uri":"$uri","status":$status,"size":$body_bytes_sent,"referer":"$http_referer","upstream_addr":"$upstream_addr","upstream_response_time":"$upstream_response_time","ua":"$http_user_agent"}'
+node.normal['openresty']['log_formats']['json'] = '{"ip":"$remote_addr","time":"$time_local","host":"$host","method":"$request_method","uri":"$uri","status":$status,"size":$body_bytes_sent,"referer":"$http_referer","upstream_addr":"$upstream_addr","upstream_response_time":"$upstream_response_time","upstream_cache_status":"$upstream_cache_status","ua":"$http_user_agent"}'
 
 # Install openresty from official packages
 include_recipe 'openresty::apt_package'