& logoagentkitdocs
DevelopersMCPTools

2 min read

Ask Question

Ask a natural-language question and get a synthesized identity answer.

What It Does

ask_question answers a free-form question from the authenticated andee's identity context. It can also include other andees as participants, using visibility-filtered facts for each participant.

Thanks for your feedback!

The MCP returns only synthesized text. Raw signal rows do not cross the MCP boundary.

Thanks for your feedback!

When To Use It

Use this for read-style personalization: preferences, routines, context about the caller, or questions involving connected/public participants.

Thanks for your feedback!

Required Scope

ScopePurpose
signals:askSynthesize answers from caller and participant identity context.

Inputs

InputRequiredDescription
questionYesFree-form natural-language question, 1-1000 characters.
participantsNoOther andees the question is about. Each entry is an &tag or a first name resolvable against the caller's active connections. Max 10 participants, each 1-64 characters.

The caller is always implicit and must not be included in participants. If a first name resolves to multiple connections, the server returns a disambiguation prompt instead of an answer.

Thanks for your feedback!

Try Saying

  • "Ask And what coffee order I usually prefer."
  • Thanks for your feedback!

  • "Ask And what Jane and I have in common."
  • Thanks for your feedback!

  • "Ask And what I should bring to dinner with Aaron."
  • Thanks for your feedback!

Invocation Example

{
  "question": "What do Aaron and Sam have in common?",
  "participants": ["Aaron", "Sam"]
}

Notes

  • Use &tag participants when possible because they are unambiguous.
  • Thanks for your feedback!

  • Use find_andees first when the participant is not already connected and the user did not provide an exact &tag.
  • Thanks for your feedback!

On this page