The references or IDs of one or more documents whose content you want to use as an example of the type of content the agent must return (you cannot mix references and IDs). Separate multiple references or IDs with plus symbols (+). There must be no space before or after a plus symbol.

Alternatively, you can set the Mode parameter to Token and use the PositiveDocs parameter to specify a state token to use to train agents.


Use the Mode parameter to specify whether PositiveDocs contains references, IDs, or a state token.

If you use document references that contain spaces or special characters, you must percent-encode these references twice to prevent the spaces from being treated as separators.

For example, encode the following document:

C:\Docs\Remote Connections to Secure Unix or Linux Machines.doc



For the AgentEdit action, by default the new documents are added to the existing training. Set ClearTraining to True to replace the existing training documents.

Actions: AgentAdd
Type: String
Example: PositiveDocs=123456789+234567891+345678912
See Also: Mode