Structured questions & UI

When the model is unsure, you want it to ask rather than guess — ideally with typed options instead of free-form back-and-forth. Pi’s extension API can replace the editor with custom UI, and several packages use this to add structured Q&A and other interactive widgets.

Alternatives

1. @juicesharp/rpiv-ask-user-question

A structured questionnaire the model can present when it would otherwise guess — typed options (single/multi-select, etc.) instead of free-form replies.

pi install npm:@juicesharp/rpiv-ask-user-question

Benefits

• Turns ambiguous moments into quick, clickable decisions.
• Reduces wrong-assumption rework.
• Part of the cohesive rpiv-* family.

Drawbacks

• Only helps if the model actually invokes it — guide it via AGENTS.md.
• Adds a tool the model must learn to reach for.

2. @juicesharp/rpiv-btw

Adds a /btw slash command for putting a one-off side question to the same model without polluting the main conversation — handy when you want a quick aside that won’t derail context.

pi install npm:@juicesharp/rpiv-btw

Benefits

• Keeps tangents out of the primary thread.
• Cheap, focused interaction.

Drawbacks

• It’s about you asking Pi, not Pi asking you (the inverse of the above).
• Single-purpose.

3. @juicesharp/rpiv-advisor

Lets the model request a second opinion from a stronger reviewer model before it acts — an interactive “should I do this?” check built into the workflow.

pi install npm:@juicesharp/rpiv-advisor

Benefits

• Catches mistakes by escalating to a stronger model at key moments.
• Improves quality on risky changes.

Drawbacks

• Extra latency and token cost per consultation.
• Requires access to a second, stronger model.

Which should I pick?

• Want Pi to ask you clear, typed questions: rpiv-ask-user-question.
• Want to ask Pi quick asides without derailing context: rpiv-btw.
• Want a stronger model to sanity-check risky actions: rpiv-advisor.

Tip

These pair well: combine rpiv-ask-user-question (resolve ambiguity up front) with rpiv-advisor (review before risky actions) for a tighter feedback loop.