This is a plugin for spamassassin.

http://spamassassin.apache.org/

It provides 2 functions that you can do eval tests on, these are

eval:public_check_to_address()

This just looks for a list of addresses that are known to be potential
spam traps, such as email addresses you might have used in mailing lists
or on usenet.

A user can set multiple public_address settings, like so:

public_address info@example.com
public_address public_email@example.com

This will check to see if the email was to one of these, use it with 
something like this:

  head PUBLIC_ADDRESS eval:public_check_to_address()
  score PUBLIC_ADDRESS 2
  describe PUBLIC_ADDRESS It was sent to a public email address.

To give email sent to one of these addresses a higher score.

The other test is slightly more complex and requires some fiddling:

eval:public_check_references('[strict]')

This will check a storage module for known message-ID's, the idea
is:

You send someone an email, your SMTP server logs the message-ID and 
the email you sent it from. We'll call this: <M100@example.com>

Later on, someone responds to your email. If they are using a modern
email client with a "Reply" function, it'll set this header:

In-Reply-To: <M100@example.com>

This is typically how a mail client does threading.

eval:check_references() will then look at it's own storage module
for <M100@example.com> if found, the test passes.

Therefore, the public_check_references() is only useful to 
alert you of NON-SPAM. You would give the test a NEGATIVE
value:

The optional 'strict' parameter (if set) tells it to ALSO confirm 
the message ID is from the same email address. This is optional because
it may not be possible/reliable to get the information from your mail
system. 

The idea is to set public addresses to a higher value, but then take them down
to a lower value if the second test passes. You might want to put
instructions in your signature alerting the recipient of this fact.

head REFERENCE_MINE eval:public_check_references('[strict]')
score REFERENCE_MINE -2
describe REFERENCE_MINE Probably a response to one of my emails.


Getting this to work is a little tricky, it needs the
configuration value: 

public_store_module 

Set to a storage implementation, included is a flatfile based
implementation and an SQL based one. 

The SQL based storage method needs these settings:

public_sql_password
  The password to connect to the database.

public_sql_username
   The username required to connect.

public_sql_dsn
   The database DSN, (See perldoc DBI)

Additionally, it needs a table:

   CREATE TABLE mid_history (
               username varchar(100) NOT NULL default '',
               email varchar(200) NOT NULL default '',
               message_id   varchar(127) NOT NULL,
               PRIMARY KEY  (username,message_id)
         ) TYPE=MyISAM;


Pretty much as the other SQL stuff works. perldoc PublicCheck::MidStoreSql
shows you the settings to change the SQL. 

There is a FlatStore module, but it's usefulness is limited to testing.
(Wouldn't want to scan a text file over and over for each email address, for
each email)

The catch here, is that you need to store all the outbound 
message ID's some place. 

I personally use an exim system filter, and scan the leading From_ line. I know
there are probably better ways. Included is an insert_mid.pl script, which
can be used as a crude filter for out-bound email, it needs a leading From_
line in order to detect the username. You should season it to taste.

Haven't done it yet.. but there should be a way to get an NNTP server to
store it's Message-ID's in a database.

There are probably a few bugs...

Copywright (C) 2005 Jamie Hoglund.

You're free to use it, same as SpamAssassin.

I'm always available for custom programming, see 
http://www.geniegate.com/ for contact information.