@pi-vault/pi-questionnaire

Stop guessing and ask better follow-up questions. @pi-vault/pi-questionnaire adds a questionnaire tool to Pi so an agent can collect several related answers in one focused interaction.

Use it when you want the agent to pause, ask a few structured questions, and then continue with clearer input.

Install

pi install npm:@pi-vault/pi-questionnaire

Then reload Pi:

/reload

Good uses

Use questionnaire for things like:

• narrowing scope before implementation
• choosing between a few approaches
• collecting rollout or release preferences
• confirming priorities before the agent starts work
• batching several related decisions into one prompt

If the user is already talking through the problem in free-form, normal conversation is usually better.

What it feels like to use

The user sees one question at a time in an interactive terminal flow, then reviews everything before submitting.

A question can lead to any of these outcomes:

• pick one option
• pick several options
• type a custom answer
• say “Chat about this” instead
• add a note with extra context

That makes it useful when you want structure without forcing every answer into a rigid preset.

Typical example

A release-planning questionnaire might look like this:

questionnaire({
  questions: [
    {
      type: "single-choice",
      id: "scope",
      header: "Scope",
      prompt: "Which scope should we ship first?",
      options: [
        {
          value: "small",
          label: "Small",
          description: "Minimal release scope",
        },
        { value: "full", label: "Full", description: "Broader first release" },
      ],
      recommendation: "small",
      allowOther: true,
      allowChat: true,
    },
    {
      type: "multi-choice",
      id: "checks",
      header: "Checks",
      prompt: "Which release checks should we run?",
      options: [
        { value: "lint", label: "Lint" },
        { value: "types", label: "Typecheck" },
        { value: "tests", label: "Tests" },
      ],
      recommendation: ["lint", "types", "tests"],
      allowChat: true,
    },
  ],
});

What the agent gets back

After submission, the tool returns a compact summary the agent can use immediately.

Example:

Scope: user selected: 1. Small
Scope note: "keep the first release tight"
Checks: user selected: 1. Lint, 2. Typecheck, 3. Tests

It can also capture answers like:

Scope: user wrote: "Ship only the core questionnaire flow"
Features: user wants to discuss this question

Writing better questionnaires

A few habits make the tool work better:

• keep it short
• ask only questions that affect the next decision
• offer clear, concrete options
• include a recommended option when you have one
• leave allowOther on when a custom answer might matter
• leave allowChat on when the user may need to talk something through

License

MIT