Active · amplifier-qna-bundle

Requirements
by conversation.

A structured Q&A system that turns a messy, freeform requirements chat into a polished, stakeholder-ready document — one question at a time.

Amplifier Community · February 2026

The Problem

Requirements gathering
is consistently broken.

💬

Freeform & Fragmented

Engineers and PMs collect requirements through Slack threads, scattered docs, and impromptu calls — with no consistent structure.

📋

Inconsistent Outputs

Every team member produces a different document. Critical fields are missing. Stakeholders can’t compare or approve them at scale.

🔄

Endless Revision Loops

Incomplete requirements surface late — during review or after development starts — forcing costly rework and delayed deliveries.

The root cause isn’t effort — it’s the absence of a structured, guided process.

The Solution

A guided conversation
that never dumps a questionnaire.

1

Select a template — the agent picks the right one for your situation (software feature, bug report, kickoff, user story).

2

Conversational Q&A — one or two questions at a time, with confirmation after complex answers. Never a wall of text.

3

Capture what wasn’t asked — organic insights get recorded as freeform notes alongside structured answers.

4

Generate a polished document — Markdown for stakeholders, JSON for systems. Review, revise, ship.

Live conversation

requirements-analyst

Let’s start with the core problem. In one or two sentences — what pain does this feature solve for the user?

Users can’t track which team reviewed a PR. They have to dig through Slack.

requirements-analyst

Got it — visibility into reviewer status. Who are the primary users: individual contributors, team leads, or both?

Mostly ICs, but leads need a summary view too.

requirements-analyst

Noted — I’ll capture a lead summary view as an implicit requirement. Does this feature replace an existing workflow or extend one?

One question at a time. Stateful. Never overwhelming.

Architecture

Two components.
One seamless experience.

Agent

requirements-analyst

A purpose-built AI persona. Selects templates, paces questions conversationally, listens for implicit requirements, confirms understanding, and presents the final document for review with revision support.

Python Tool · 7 operations

tool-qna

The stateful engine. Manages sessions, validates typed answers, tracks conditional question logic, records freeform notes, reports progress, and generates dual-format output.

4 Built-in Templates

⚙️

software_requirements

4 sections · ~19 questions

📚

user_story

3 sections · ~11 questions

🐛

bug_report

4 sections · ~15 questions

🚀

project_kickoff

4 sections · ~16 questions

Standout Features

Intelligence built
into the conversation.

🌿

Conditional Branching

Questions unlock dynamically based on prior answers. Say “yes” to authentication requirements, and the session reveals auth-specific follow-ups — automatically.

ADAPTIVE FLOW

📝

Freeform Notes

Real conversations surface insights no template anticipates. The record_note operation captures organic context alongside structured answers — so nothing valuable is lost.

NOTHING LEFT BEHIND

✅

Type-Aware Validation

Answers are validated against their expected type: yes_no, choice, multi_choice, scale, or text. No silent bad data.

CLEAN DATA

🧭

Per-Section Coaching

Each template section includes LLM coaching notes — guidance on tone, depth, when to push back, and what signals to watch for. Context for the agent, consistency for the conversation.

GUIDED INTELLIGENCE

⏭️

Graceful Skip Support

Optional questions can be skipped without breaking the session flow. The agent acknowledges the skip and moves forward — no awkward dead ends.

USER-FRIENDLY

📊

Live Progress Tracking

The get_progress operation returns section-by-section completion status at any point. Users always know where they are and what’s left.

ALWAYS ORIENTED

Extensibility & Composability

Zero code to extend.
One behavior to compose.

📁 Custom Templates via YAML

Drop a .yaml file into the templates/ directory. No code changes, no rebuilds. The system picks it up automatically. Demonstrated out-of-the-box by api-design-review.yaml.

🧩 Composable Behavior

The qna behavior is a standalone Amplifier module. Any bundle can include it in a single line — adding structured Q&A to workflows that weren’t designed for it from the start.

📄 Markdown — stakeholder-ready

{} JSON — machine-readable

# A custom template — zero Python required

id: api_design_review
name: API Design Review
description: Structured review for REST API proposals

sections:
  - id: fundamentals
    title: API Fundamentals
    coaching: |
        Focus on resource naming and HTTP semantics.
        Push back on verbs in paths.
    questions:
      - id: rest_maturity
        text: What REST maturity level are you targeting?
        type: choice
        options: [Level 0, Level 1, Level 2, Level 3]
        required: true

      - id: auth_scheme
        text: What authentication scheme?
        type: choice
        condition:
          question_id: rest_maturity
          operator: in
          value: [Level 2, Level 3]
          # Only shown for higher maturity levels

Sources & Methodology

What this deck
is built on.

Primary source: Comprehensive bundle specification provided by the bundle author, covering all components, operations, templates, and design decisions.

Bundle repository: amplifier-qna-bundle — reviewed for architecture, component inventory, and template catalog.

Template inventory: 4 built-in templates confirmed: software_requirements, user_story, bug_report, project_kickoff. Plus example custom template: api-design-review.yaml.

Operation inventory: 7 tool-qna operations verified: list_templates, start_session, get_questions, record_answer, record_note, get_progress, generate_report.

Question counts: Approximate (~) figures used as provided. Exact counts may vary with conditional question expansion at runtime.

Chat example on slide 3 is illustrative — constructed to demonstrate the conversational pacing pattern, not a recorded session.

Data as of: February 27, 2026 · Feature status: Active · Primary contributor: bundle author

Get Started

Stop gathering requirements.
Start having conversations.

Clone the bundle, start a session, and have your first stakeholder-ready document in one conversation.

📦

Clone the Bundle

Get the bundle and its four built-in templates.

amplifier-qna-bundle

🧩

Compose It

Add the qna behavior to any Amplifier bundle you already own.

behaviors: [qna]

✍️

Extend with YAML

Write a custom template for your domain — no code, just YAML.

templates/my-template.yaml

amplifier-qna-bundle · Active · Amplifier Community

Requirementsby conversation.

Requirements gatheringis consistently broken.

Freeform & Fragmented

Inconsistent Outputs

Endless Revision Loops

A guided conversationthat never dumps a questionnaire.

Two components.One seamless experience.

Intelligence builtinto the conversation.

Conditional Branching

Freeform Notes

Type-Aware Validation

Per-Section Coaching

Graceful Skip Support

Live Progress Tracking

Zero code to extend.One behavior to compose.

📁 Custom Templates via YAML

🧩 Composable Behavior

What this deckis built on.

Stop gathering requirements.Start having conversations.

Clone the Bundle

Compose It

Extend with YAML

Requirements
by conversation.

Requirements gathering
is consistently broken.

A guided conversation
that never dumps a questionnaire.

Two components.
One seamless experience.

Intelligence built
into the conversation.

Zero code to extend.
One behavior to compose.

What this deck
is built on.

Stop gathering requirements.
Start having conversations.