Label your questions

Group questions into categories and mark which ones ask for a judgment - so verdicts are counted where they make sense.

For: Anyone whose project has grown past a list you can read at a glance

Paste this into your assistant

Look at our tracked questions. Which ones ask for a list of names rather than one pick? Mark those as not needing a verdict, and put all of them into short categories.

What happens

1. 01Reads the tracked questions with their current labels.
2. 02Proposes categories and which questions expect a single judgment, and shows you the change first. Nothing changes until you confirm.

Tools it calls

• listPromptsread only

  The questions being tracked for a project, with their intent and topic tags. Read this before proposing new prompts - a duplicate takes a prompt slot from the plan and measures the same question twice, every day, for nothing.

• updatePromptasks before it acts

  Set the category, framing (neutral, partisan, branded) or expectsVerdict of one or many tracked prompts. expectsVerdict decides which questions get a verdict check. Values set here are fixed - the model will not relabel them. Returns a plan first; call again with confirm to apply.

What you end up with

Labelled questions, and verdicts counted only where a question asks for one.

Answer analysis switched on for the project - it reads every answer with a model, so the Scorra team turns it on per project. Ask us. MCP is included from Growth up. How to connect.