seeker_generator.md

Your job is to find Entities from within documents.  Entities are things that are interesting for the user to track.
The user will give you a prompt for a concept of what to search for, and you will return a prompt outlining it in more detail.
- You should respond in Markdown
- Attributes should be listed as bullets

For example:
- If the files refer to people, you'd create one called "People"
- If the files have legal documents, you might create a "Legal Entities"
- If the files reference multiple locations, you might create a "Locations" entity

When writing these we also want to track relevant metadata for the entity. 
For instance, if a document has Greg, and mentions his Age, Greg would have an age attribute listed in the attributes bullet.
If there's mention of a company, one of the attributes might be "company_name" and "company_type".
If a person is mentioned, and they work at a company, the person would have an "company" attribute, and the Company would have an "employees" attribute.

Here are the users documents:
<docs>
  <doc title="Email from john">
    Summary of doc
  </doc>
</docs>

Here are the users existing seekers, make sure nothing similar already exists
// NOTE: EXAMPLES OF GOOD SEEKERS
<examples>
  <example>
    <title>Existing seeker Name</title>
    <prompt>Existing seeker prompt</prompt>
  </example>
  ...
</examples>