This information collator uses scripted agents to retrieve web data.

Each parser is recorded from an actual HTTP session with HTTP::Recorder.

Initially I intended to load user data from several sites:

 - cpan
 - foafweb
 - friendster
 - orkut
 - rrn (phone book)

TODO: Karma lookup. Notes fields are often manual-entry reputation trackers.

Each of the services provides an address book associated with a unique address.

 - Most address books have names and/or email address.
 - Field names are not standardized.  (URL, Site, Homepage, Link, Blog.)
 - Merge data from multiple address books using shared data.  (Name, Email.)

Additional services TODO list:

 - gpg keyserver (http, ldap)
 - social network links (friendster, opml)
 - dns records (srv, mx, txt; spf, domainkeys)

Potential use:

1. Fetch the list of Orkut friends' email addresses.

2. Lookup each email address in the keyservers.

3. Query the DNS SRV records for each key.

4. Poll the stub servers for waiting mail.
   The client signs each request with a private key.

5. The server checks the signature against the local keyrings.
   When it matches, the server presents a list of waiting stubs.

6. The client compares its keyring to the list of stubs.
   When the fingerprints match, the client fetches the message.