独家

Lifelines `<mv-lifelines>`

为艰难的单项选择（方案、配送方式、笔记本配置、保险档位、实例规格）提供决策辅助，每个辅助都是一次求助机会，使用次数有限，用掉后清晰可见。

文化出处

谁想成为百万富翁？，David Briggs、Mike Whitehill 和 Steven Knight（ITV）（1998 年，电视问答节目）。卡住的参赛者每种求助机会都可使用一次：一种去掉两个错误答案，一种显示现场观众的投票结果，一种打电话给信任的朋友。在 UI 中，同样的三种辅助手段帮助做出艰难的产品选择：搁置不合适的选项（附理由），显示相似客户的选择，并把候选清单发给同事投票。每种用过后都会明显标示为已用，并由一个信心读数将它们综合起来。

工作原理

针对艰难单选（方案、配送选项、笔记本电脑配置、保险档位、实例规格）的决策辅助工具，每个都是一次使用次数有限、用过后会明显标示的求助机会。它包裹你自己的原生单选按钮或选项卡片，从不替换它们，因此表单值、校验和方向键仍由浏览器处理。“Narrow it down” 会利用你的 narrow(context) 函数或 data-fit 评分，把明显不符合用户所述需求的选项搁置一旁。每次移除都附带理由，当前选择从不会被隐藏，会保留最少数量的选项，而 “Show them” 会把被搁置的选项重新显示出来，依然可选。“What others chose” 会为每个选项添加带百分比的分布条，包括整体数据和 “people like you” 细分人群的数据，来源可以是 data-share 特性、统计对象或函数，并标出被选最多的选项。“Ask someone” 会撰写一条可编辑的消息，包含候选清单和你当前的选择，以及可选的投票链接，并通过 Web Share API 发送或复制。回复通过 addVote() 回传，并显示在每个选项上。信心读数将当前选择的匹配度、大众选择和投票综合为 Strong / Moderate / Low 等级并附理由。每种求助都会经过可取消的 mv-lifeline 事件，其 waitUntil(promise) 可从你的服务器获取匹配度、统计数据或投票链接。

分类	表单
类型	Web Component（`<mv-lifelines>`）
状态	稳定版
同时安装	button, textarea
Keywords	exclusive, culture, decision, choice, radio-group, option-cards, plan-picker, pricing, shipping, configurator, recommendation, social-proof, share, web-share, voting, confidence, filter, form

When to use

Users stall on a plan, tier or configuration picker and need help ruling out options before they commit
The app knows the user's needs (team size, deadline, budget) and can explain why an option doesn't fit
Showing what similar customers picked would reassure a hesitant buyer without pushing a single recommendation
A purchase is often decided with a colleague or partner, and the shortlist should be easy to send them for a vote

Avoid when

The choice is between two to five quick, low-stakes settings like view mode or alignment → use Segmented instead
The options only need to be listed as cards with a price and badge, with no decision help → use Radio instead
Plans must be compared feature by feature in a full pricing section with a billing toggle → use Pricing instead

安装

node scripts/add.mjs lifelines --out ./src/marvelous

使用 Marvelous UI MCP 服务器的 AI 智能体：install_components({ slugs: ["lifelines"], target_dir: "<absolute path>/src/marvelous", framework: "react" })。

复制的文件（含依赖）：tokens/tokens.css, core/base.css, components/button/button.css, core/dom.js, core/element.js, components/textarea/textarea.css, components/textarea/char-count.js, core/motion.js, components/lifelines/lifelines.js, components/lifelines/lifelines.css。

用法

标准标记，可在此基础上通过属性、data-* 和 CSS 变量进行定制：

<div id="ll-demo" style="width:min(100%,66rem);margin-inline:auto">
  <style>
    #ll-demo { display:grid; gap:1.25rem; align-content:start }
    #ll-demo .ll-layout { display:grid; grid-template-columns:minmax(0,1fr) 16.5rem; gap:1.25rem; align-items:start }
    #ll-demo .ll-card { min-width:0; border:1px solid var(--mv-border); border-radius:var(--mv-radius-xl); background:var(--mv-surface); box-shadow:var(--mv-shadow-sm) }
    #ll-demo .ll-head { display:flex; align-items:flex-end; justify-content:space-between; gap:.75rem 1.25rem; flex-wrap:wrap; padding:1.125rem 1.25rem 1rem; border-bottom:1px solid var(--mv-border) }
    #ll-demo .ll-head h3 { margin:0; font-size:1.0625rem; letter-spacing:-.01em }
    #ll-demo .ll-head p { margin:.1875rem 0 0; color:var(--mv-fg-muted); font-size:.8125rem }
    #ll-demo .ll-needs { display:flex; align-items:flex-end; gap:.75rem; flex-wrap:wrap }
    #ll-demo .ll-need { display:grid; gap:.3125rem; font-size:.75rem; font-weight:500; color:var(--mv-fg-muted) }
    #ll-demo .ll-need .mv-select { width:auto; min-width:7.5rem; height:var(--mv-control-sm); font-size:.8125rem }
    #ll-demo .ll-need .mv-choice { height:var(--mv-control-sm); align-items:center; font-size:.8125rem }
    #ll-demo .ll-body { padding:1.125rem 1.25rem 1.25rem }
    #ll-demo .mv-choice-group legend { font-size:.8125rem; color:var(--mv-fg-muted) }
    #ll-demo .mv-radio-meta { margin-inline-start:auto; padding-inline-start:.5rem; font-weight:600; font-size:.875rem; white-space:nowrap; font-variant-numeric:tabular-nums }
    #ll-demo .mv-radio-meta small { color:var(--mv-fg-muted); font-weight:500; font-size:.75rem }
    #ll-demo .ll-side { position:sticky; top:1rem; display:grid; gap:1rem; padding:1.125rem; border:1px solid var(--mv-border); border-radius:var(--mv-radius-xl); background:var(--mv-surface-raised); box-shadow:var(--mv-shadow-sm); font-size:.8125rem }
    #ll-demo .ll-side h4 { display:flex; align-items:center; justify-content:space-between; gap:.5rem; margin:0; font-size:.8125rem; font-weight:600 }
    #ll-demo .ll-side p { margin:0; color:var(--mv-fg-muted); font-size:.75rem; line-height:1.45 }
    #ll-demo .ll-votes { display:grid; gap:.5rem }
    #ll-demo .ll-votes .mv-button { justify-content:flex-start; width:100% }
    #ll-demo .ll-sep { height:1px; background:var(--mv-border) }
    #ll-demo .ll-log { display:grid; gap:.3125rem; margin:0; padding:0; list-style:none; font:.6875rem/1.45 var(--mv-font-mono); color:var(--mv-fg-muted) }
    #ll-demo .ll-log li { overflow:hidden; text-overflow:ellipsis; white-space:nowrap }
    #ll-demo .ll-log b { color:var(--mv-fg); font-weight:500 }
    #ll-demo .ll-ship .ll-body { display:grid; grid-template-columns:minmax(0,15rem) minmax(0,1fr); gap:1.5rem; align-items:start }
    #ll-demo .ll-order { display:grid; gap:.625rem; font-size:.8125rem }
    #ll-demo .ll-order dl { display:grid; gap:.375rem; margin:0 }
    #ll-demo .ll-order dl div { display:flex; justify-content:space-between; gap:.5rem }
    #ll-demo .ll-order dt { color:var(--mv-fg-muted) }
    #ll-demo .ll-order dd { margin:0; font-variant-numeric:tabular-nums }
    #ll-demo .ll-order .ll-by { display:flex; gap:.5rem; align-items:flex-start; padding:.625rem .75rem; border-radius:var(--mv-radius-lg); background:var(--mv-bg-subtle); color:var(--mv-fg-muted); font-size:.75rem; line-height:1.45 }
    #ll-demo .ll-order .ll-by b { color:var(--mv-fg); font-weight:600 }
    #ll-demo .ll-hint { margin:0; color:var(--mv-fg-muted); font-size:.75rem; text-align:center }
    @media (max-width:54rem) {
      #ll-demo .ll-layout { grid-template-columns:minmax(0,1fr) }
      #ll-demo .ll-side { position:static }
      #ll-demo .ll-ship .ll-body { grid-template-columns:minmax(0,1fr) }
    }
  </style>

  <div class="ll-layout">
    <!-- Main: choosing a workspace plan, with the app's own narrow() and stats() -->
    <section class="ll-card" aria-labelledby="ll-plan-title">
      <header class="ll-head">
        <div>
          <h3 id="ll-plan-title">Choose a plan for Kestrel Studio</h3>
          <p>You can switch plans at any time. Prices billed yearly.</p>
        </div>
        <div class="ll-needs" role="group" aria-label="Your needs">
          <label class="ll-need">Team size
            <select class="mv-select" id="ll-seats">
              <option value="2">1-3 people</option>
              <option value="8" selected>4-10 people</option>
              <option value="18">11-25 people</option>
              <option value="60">26-100 people</option>
            </select>
          </label>
          <label class="ll-need">Storage
            <select class="mv-select" id="ll-storage">
              <option value="5">5 GB</option>
              <option value="50" selected>50 GB</option>
              <option value="500">500 GB</option>
            </select>
          </label>
          <span class="ll-need">Security
            <label class="mv-choice"><input type="checkbox" role="switch" class="mv-switch" id="ll-sso"> SSO required</label>
          </span>
        </div>
      </header>
      <div class="ll-body">
        <mv-lifelines id="ll-plan" similar-label="teams of 4-10"
          question="Which plan should we get for the design team? We’re 8 people."
          share-url="https://app.example.com/plans/vote?options={shortlist}&amp;pick={pick}">
          <fieldset class="mv-choice-group" data-layout="grid" style="--mv-choice-min:17rem">
            <legend>Plan</legend>
            <label class="mv-choice-card">
              <input type="radio" class="mv-radio" name="ll-plan" value="starter">
              <span class="mv-choice-text"><span class="mv-choice-title">Starter</span><span class="mv-choice-description">Up to 3 seats · 5 GB · Community support</span></span>
              <span class="mv-radio-meta">$0</span>
            </label>
            <label class="mv-choice-card">
              <input type="radio" class="mv-radio" name="ll-plan" value="team" checked>
              <span class="mv-choice-text"><span class="mv-choice-title">Team</span><span class="mv-choice-description">Up to 25 seats · 100 GB · Shared libraries</span></span>
              <span class="mv-radio-meta">$12 <small>/ seat</small></span>
            </label>
            <label class="mv-choice-card">
              <input type="radio" class="mv-radio" name="ll-plan" value="business">
              <span class="mv-choice-text"><span class="mv-choice-title">Business</span><span class="mv-choice-description">Unlimited seats · 1 TB · SSO and audit log</span></span>
              <span class="mv-radio-meta">$24 <small>/ seat</small></span>
            </label>
            <label class="mv-choice-card">
              <input type="radio" class="mv-radio" name="ll-plan" value="enterprise">
              <span class="mv-choice-text"><span class="mv-choice-title">Enterprise</span><span class="mv-choice-description">From 50 seats · SSO and SCIM · 99.99% SLA</span></span>
              <span class="mv-radio-meta">Custom</span>
            </label>
          </fieldset>
        </mv-lifelines>
      </div>
    </section>

    <aside class="ll-side" aria-label="Demo controls">
      <h4>Replies you received <span class="mv-badge" data-variant="secondary" data-shape="pill" id="ll-count">0 votes</span></h4>
      <p>Your app collects answers from the shared link and feeds each one back with <code>addVote()</code>.</p>
      <div class="ll-votes">
        <button class="mv-button" data-variant="outline" data-size="sm" type="button" data-vote="team" data-from="Priya Raman" data-note="Team covers us, keep the budget for fonts.">Priya picks Team</button>
        <button class="mv-button" data-variant="outline" data-size="sm" type="button" data-vote="business" data-from="Marcus Olsen" data-note="We’ll want SSO once the agency joins.">Marcus picks Business</button>
        <button class="mv-button" data-variant="outline" data-size="sm" type="button" data-vote="team" data-from="Aiko Tanaka">Aiko picks Team</button>
      </div>
      <div class="ll-sep"></div>
      <button class="mv-button" data-variant="secondary" data-size="sm" type="button" id="ll-reset">Give lifelines back</button>
      <div class="ll-sep"></div>
      <h4>Events</h4>
      <ol class="ll-log" id="ll-log"><li>-</li></ol>
      <span class="mv-sr-only" role="status" id="ll-status"></span>
    </aside>
  </div>

  <!-- Variant: declarative data only (data-fit, data-share), two lifelines, no confidence readout -->
  <section class="ll-card ll-ship" aria-labelledby="ll-ship-title">
    <header class="ll-head">
      <div>
        <h3 id="ll-ship-title">Shipping</h3>
        <p>Order #48213 · 2 items · to Chicago, IL 60614</p>
      </div>
    </header>
    <div class="ll-body">
      <div class="ll-order">
        <dl>
          <div><dt>Trail runner, size 9</dt><dd>$128.00</dd></div>
          <div><dt>Merino socks, 2 pairs</dt><dd>$24.00</dd></div>
        </dl>
        <p class="ll-by"><span>You said it’s a gift for <b>Friday, October 2</b>. Options that won’t arrive in time can be set aside.</span></p>
      </div>
      <mv-lifelines lifelines="narrow, poll" threshold="0.7" show-confidence="false" similar-label="shoppers in Chicago" poll-note="Orders shipped to Illinois in the last 30 days." label="Not sure which one?">
        <fieldset class="mv-choice-group">
          <legend>Delivery</legend>
          <label class="mv-choice-card" data-fit="0" data-fit-reason="Arrives Oct 5-7, after Friday" data-share="38" data-share-similar="22">
            <input type="radio" class="mv-radio" name="ll-ship" value="economy">
            <span class="mv-choice-text"><span class="mv-choice-title">Economy</span><span class="mv-choice-description">6-8 business days</span></span>
            <span class="mv-radio-meta">Free</span>
          </label>
          <label class="mv-choice-card" data-fit="0.55" data-fit-reason="Arrives Sep 30 to Oct 2, cutting it close" data-share="41" data-share-similar="44">
            <input type="radio" class="mv-radio" name="ll-ship" value="standard" checked>
            <span class="mv-choice-text"><span class="mv-choice-title">Standard</span><span class="mv-choice-description">3-5 business days</span></span>
            <span class="mv-radio-meta">$6.99</span>
          </label>
          <label class="mv-choice-card" data-fit="1" data-fit-reason="Arrives Monday, Sep 28" data-share="16" data-share-similar="27">
            <input type="radio" class="mv-radio" name="ll-ship" value="express">
            <span class="mv-choice-text"><span class="mv-choice-title">Express</span><span class="mv-choice-description">2 business days, tracked</span></span>
            <span class="mv-radio-meta">$14.99</span>
          </label>
          <label class="mv-choice-card" data-fit="0.85" data-fit-reason="Ready Thursday, Oct 1 at Lakeview Market" data-share="5" data-share-similar="7">
            <input type="radio" class="mv-radio" name="ll-ship" value="pickup">
            <span class="mv-choice-text"><span class="mv-choice-title">Pickup point</span><span class="mv-choice-description">Collect from a locker near you</span></span>
            <span class="mv-radio-meta">$3.99</span>
          </label>
        </fieldset>
      </mv-lifelines>
    </div>
  </section>
  <p class="ll-hint">Each lifeline works once · “Narrow it down” was already used on the plan picker · change your needs to re-check the set-aside plans</p>

  <script type="module">
    await customElements.whenDefined("mv-lifelines");
    const $ = (id) => document.getElementById(id);
    const ll = $("ll-plan");
    const PLANS = {
      starter: { seats: 3, storage: 5, sso: false, min: 1 },
      team: { seats: 25, storage: 100, sso: false, min: 1 },
      business: { seats: Infinity, storage: 1000, sso: true, min: 1 },
      enterprise: { seats: Infinity, storage: 5000, sso: true, min: 50 },
    };
    const readNeeds = () => ({ seats: Number($("ll-seats").value), storage: Number($("ll-storage").value), sso: $("ll-sso").checked });
    const gb = (n) => (n >= 1000 ? `${n / 1000} TB` : `${n} GB`);

    // The app's own fit logic: 0..1 per option, with the reason shown to the user.
    ll.needs = readNeeds();
    ll.narrow = ({ needs, options }) => options.map(({ value }) => {
      const p = PLANS[value];
      if (p.seats < needs.seats) return { value, fit: 0.1, reason: `Up to ${p.seats} seats, you need ${needs.seats}` };
      if (p.min > needs.seats) return { value, fit: 0.2, reason: `Starts at ${p.min} seats, you need ${needs.seats}` };
      if (p.storage < needs.storage) return { value, fit: 0.15, reason: `${gb(p.storage)} of storage, you need ${gb(needs.storage)}` };
      if (needs.sso && !p.sso) return { value, fit: 0.15, reason: "No single sign-on" };
      if (!needs.sso && p.sso) return { value, fit: 0.65, reason: "Fits, but you’d pay for SSO you don’t need" };
      return { value, fit: 0.92, reason: `Room for ${Number.isFinite(p.seats) ? p.seats : "unlimited"} seats and ${gb(p.storage)}` };
    });
    // What others chose: overall and for teams of the same size.
    const SEGMENTS = {
      2: ["teams of 1-3", { starter: 71, team: 22, business: 6, enterprise: 1 }],
      8: ["teams of 4-10", { starter: 9, team: 58, business: 29, enterprise: 4 }],
      18: ["teams of 11-25", { starter: 2, team: 47, business: 43, enterprise: 8 }],
      60: ["teams of 26-100", { starter: 1, team: 12, business: 52, enterprise: 35 }],
    };
    ll.stats = ({ needs }) => {
      const [similarLabel, similar] = SEGMENTS[needs.seats];
      return {
        everyone: { starter: 41, team: 33, business: 19, enterprise: 7 },
        similar, similarLabel,
        note: "Based on 12,480 workspaces created in the last 90 days.",
      };
    };

    const log = $("ll-log");
    const push = (html) => {
      if (log.firstElementChild?.textContent === "-") log.replaceChildren();
      const li = document.createElement("li");
      const [name, rest] = html;
      li.append(Object.assign(document.createElement("b"), { textContent: name }), ` ${rest}`);
      log.prepend(li);
      while (log.children.length > 5) log.lastElementChild.remove();
    };
    ll.addEventListener("mv-lifeline", (e) => push(["mv-lifeline", `${e.detail.lifeline}${e.detail.refresh ? " (refresh)" : ""}`]));
    ll.addEventListener("mv-lifeline-result", (e) => push(["result", `${e.detail.lifeline} · ${e.detail.confidence?.level ?? "no pick"}`]));
    ll.addEventListener("mv-change", (e) => push(["mv-change", `${e.detail.value} · ${e.detail.confidence?.level ?? "-"}`]));
    ll.addEventListener("mv-share", (e) => push(["mv-share", e.detail.method]));
    ll.addEventListener("mv-vote", (e) => {
      push(["mv-vote", `${e.detail.vote.from} → ${e.detail.vote.value}`]);
      const n = e.detail.votes.length;
      $("ll-count").textContent = `${n} vote${n === 1 ? "" : "s"}`;
    });

    // Start with one lifeline already spent, so the mechanic is visible.
    ll.use("narrow");

    for (const id of ["ll-seats", "ll-storage", "ll-sso"]) {
      $(id).addEventListener("change", () => {
        ll.needs = readNeeds();
        ll.refresh();
      });
    }
    document.querySelectorAll("#ll-demo [data-vote]").forEach((btn) => {
      btn.addEventListener("click", () => {
        ll.addVote({ value: btn.dataset.vote, from: btn.dataset.from, note: btn.dataset.note });
        btn.disabled = true;
      });
    });
    $("ll-reset").addEventListener("click", async () => {
      await ll.reset();
      document.querySelectorAll("#ll-demo [data-vote]").forEach((b) => { b.disabled = false; });
      $("ll-count").textContent = "0 votes";
      $("ll-status").textContent = "All lifelines are available again.";
    });
  </script>
</div>

API

Attributes

Name	类型	Default	Description
`lifelines`	comma list of narrow \| poll \| ask	`narrow, poll, ask`	Which lifelines are offered, in toolbar order.
`uses`	number \| "unlimited" \| "narrow:1, poll:2, ask:unlimited"	`1`	How many times each lifeline can be used: one number for all, or per lifeline ("*" sets the default for the others). Up to five uses show as dots, more as a count.
`threshold`	number (0..1 or %)	`0.5`	Fit below which “Narrow it down” sets an option aside. Changing it re-runs a narrowing already done, without spending a use.
`keep`	number	`2`	Minimum number of options left visible after narrowing. The best-fitting of the would-be removals are kept to reach it. The checked option is never set aside; if it fits poorly it is flagged “Might not fit” instead.
`name`	string		Name of the radio group to manage when the element contains several. Defaults to the name of the first radio inside.
`label`	string	`Need a hand deciding?`	Visible label of the lifelines toolbar (also its accessible name). An empty value hides it.
`similar-label`	string	`people like you`	Name of the “people like you” segment, written lowercase to read inside sentences (“Chosen by 58% of teams of 4-10”). It is capitalized where it starts a label. A stats object's similarLabel wins.
`poll-note`	string		Where the statistics come from, shown in the “What others chose” note (“Orders shipped to Illinois in the last 30 days.”). A stats object's note wins.
`share-url`	URL template		Link added to the “Ask someone” message, for example your vote page. {shortlist} (comma-separated values still visible) and {pick} are replaced, URI-encoded. A url resolved by waitUntil wins.
`question`	string	`Could you help me decide? Here are the options I’m weighing:`	First line of the “Ask someone” message and title of the Web Share payload.
`show-confidence`	"true" \| "false"	`true`	"false" hides the confidence readout (the confidence property still works).
`data-fit / data-fit-reason`	number (0..1 or %) / string		On an option (its label, a [data-option] wrapper or the radio): declarative fit score and the reason shown to the user. Used when narrow() is not set or returns nothing for that option.
`data-share / data-share-similar`	number		On an option: share of everyone and of the “people like you” segment who chose it. Percentages, fractions (sum ≤ 1) or raw counts (sum above 100) are all normalized to percentages.
`data-option / data-label / data-summary`	attribute / string / string		On an option: data-option marks the wrapper when it isn't the radio's `<label>`; data-label and data-summary override the name and summary used in announcements and in the “Ask someone” message (defaults: .mv-choice-title, then .mv-choice-description · .mv-radio-meta).
`data-lifelines="toolbar \| notes \| readout"`	attribute		Place on your own empty containers to choose where the toolbar, the notes and the readout go. By default the toolbar and notes are prepended and the readout appended.
`data-lifeline-fit / data-lifeline-top / data-lifelines-hidden`	set by the component		On each option: fit \| poor \| out after narrowing, data-lifeline-top on the most chosen option, data-lifelines-hidden while an option is set aside and hidden (styleable).

Properties

Name	类型	Description
`narrow`	(context) => fits \| Promise`<fits>`	Your fit logic. context: { needs, value, options: [{ value, label, summary, element, data }], setAside }. Return [{ value, fit, reason }] or { [value]: fit \| { fit, reason } }, with fit in 0..1 (or a percentage).
`stats`	object \| (context) => object \| Promise`<object>`	What others chose: { everyone: { [value]: share }, similar?: { [value]: share }, similarLabel?, everyoneLabel?, note? }, a flat { [value]: share } map, or [{ value, share, similar }]. Setting it after the lifeline was used updates the bars in place.
`needs`	any	The user's stated needs, passed to narrow() and stats() as context.needs. Call refresh() after changing it.
`value`	string	Value of the checked radio. Setting it checks the matching radio (showing it again if it was set aside) without emitting mv-change.
`confidence`	{ value, level, label, pick, signals } \| null	Read-only. value 0..1 (null without signals), level "high" (≥ 0.7) \| "medium" (≥ 0.4) \| "low", signals: [{ kind: "fit" \| "crowd" \| "votes", value, tone: "good" \| "neutral" \| "bad", text }]. Weighted mean: fit 0.5, crowd 0.25 (the pick's share relative to the most chosen), votes 0.25.
`votes`	[{ value, from, note, at }]	Answers received through addVote() (read-only copy).
`setAside`	string[]	Values currently set aside by “Narrow it down” (read-only).
`strings`	Partial<Record<string, string>>	Overrides for every visible text and announcement (toolbarLabel, narrow, poll, ask, used, narrowTitle, narrowText, showAside, hideAside, tagOut, tagPoor, pollTitle, everyone, similar, mostChosen, mostChosenBy, askTitle, askText, share, copy, copied, question, askLeaning, askReply, askLink, answers, confidenceFor, levelHigh, levelMedium, levelLow, signalFit, signalCrowd, signalVotes, announce*…). English defaults.

Methods

Name	Description
`use(name)`	Runs a lifeline ("narrow" \| "poll" \| "ask") as if its button was pressed. Resolves to true when it ran; for ask, when its panel opened. A lifeline with nothing to show (no data) is not spent.
`usesLeft(name)`	Remaining uses of a lifeline (Infinity when unlimited).
`refresh()`	Re-runs the lifelines already used (narrow and poll) against the current needs and data, without spending a use. mv-lifeline fires with refresh: true.
`showAll(force?)`	Shows (true) or hides (false) the set-aside options; toggles without argument. Resolves to the new state.
`addVote({ value, from?, note? })`	Feeds back an answer from someone you asked. A second vote from the same person replaces the first. Returns false for an unknown value.
`clearVotes()`	Removes every vote.
`reset()`	Gives every lifeline back and clears narrowing, statistics and votes; set-aside options come back.
`openAsk() / closeAsk()`	Opens or closes the “Ask someone” panel.

Events

Name	Description
`mv-lifeline`	Cancelable, before a lifeline runs (for ask: when its panel opens for a new use). detail: { lifeline, refresh, usesLeft, context, waitUntil(promise) }. preventDefault() vetoes it (nothing is spent). Promises passed to waitUntil show a busy state; the last non-null value they resolve to is used as the data: fits for narrow, stats for poll, { text?, url? } for ask. A rejection spends nothing and emits mv-lifeline-error.
`mv-lifeline-result`	A lifeline ran. detail: { lifeline, refresh, result, confidence }. result: { setAside: [{ value, label, fit, reason }], fits } for narrow, { everyone, similar, top } for poll, { method, text, url } for ask.
`mv-lifeline-error`	A waitUntil promise, narrow() or stats() threw. detail: { lifeline, error }.
`mv-change`	The user picked another option. detail: { value, label, confidence }. The native change event still bubbles from the radio.
`mv-share`	The “Ask someone” message was shared or copied. detail: { method: "share" \| "copy", text, url }. The first one of a use spends it.
`mv-vote`	addVote() recorded an answer. detail: { vote, votes, confidence }.

Content structure

Name	Description
`(content)`	Your radio group: a `<fieldset>` of .mv-choice-card or .mv-choice labels, or any markup with native radios. Per-option results go into [data-lifelines-slot] if present, else .mv-choice-text, else the option wrapper.

CSS classes

Name	Description
`mv-lifelines-toolbar`	Labelled group holding -toolbar-label and .mv-lifelines-buttons.
`mv-lifelines-lifeline`	A lifeline button: -lifeline-icon, -lifeline-label, .mv-lifelines-uses (dots, count or “Used”). data-lifeline, data-state="available \| busy \| used", data-spent during the spend pulse.
`mv-lifelines-note / mv-lifelines-ask`	What each lifeline found, under the toolbar: -note-icon, -note-title, -note-text, -note-list, .mv-lifelines-legend. The ask panel adds .mv-lifelines-message (textarea), -ask-actions and .mv-lifelines-answers.
`mv-lifelines-aside`	Per-option results (aria-hidden, linked to the radio with aria-describedby): .mv-lifelines-tag, .mv-lifelines-reason[data-kind], .mv-lifelines-poll > .mv-lifelines-row[data-segment] (-row-label, .mv-lifelines-meter, -row-value), .mv-lifelines-votes.
`mv-lifelines-readout`	Confidence readout, data-level="high \| medium \| low \| none": -readout-title, .mv-lifelines-level, .mv-lifelines-gauge, .mv-lifelines-signals > li[data-tone].

CSS variables

Name	Default	Description
`--mv-lifelines-accent`	`var(--mv-accent)`	Lifeline icons, remaining-use dots, spend pulse and the “Moderate” level.
`--mv-lifelines-everyone`	`38% fg mixed into bg`	Bar color of the “everyone” segment.
`--mv-lifelines-similar`	`var(--mv-lifelines-accent)`	Bar color of the “people like you” segment.
`--mv-lifelines-good`	`var(--mv-success)`	Good-fit marks and the “Strong” level.
`--mv-lifelines-bad`	`var(--mv-warning)`	Poor-fit marks and the “Low” level.

Accessibility

The native radios are never replaced: the group keeps its <fieldset>/<legend>, arrow keys, required validation and form value. Lifelines are real <button>s in a group labelled by the visible toolbar label. Each button's name includes its state (“Narrow it down, 1 use left” / “, used” / “, working” with aria-busy). A spent lifeline keeps focus and gets aria-disabled="true"; pressing it announces that it was already used. The “Ask someone” button stays operable to reopen its panel, with aria-expanded and aria-controls. Results are announced once in a polite live region (“Set aside Starter and Enterprise. 2 options left. Confidence in Team: strong.”; “Most chosen by teams of 4-10: Team, 58%.”; “Priya Raman picked Team.”); selecting options never chatters. Per-option results are visible text, bars included (every bar has its percentage written next to it). They are aria-hidden inside the label so they don't lengthen the radio's name, and linked with aria-describedby, so a screen reader hears “Team, radio button, 2 of 4. Most chosen by teams of 4-10. Good fit: Room for 25 seats and 100 GB. Everyone 33%, teams of 4-10 58%”. Set-aside options leave the layout and the arrow-key cycle; the “Show them” toggle (aria-pressed, aria-controls pointing at the hidden options) brings them back selectable, marked “Set aside” with their reason, and the checked option is never hidden. The ask panel is a labelled region with a real <label> on its editable message; focus moves into it on open, and Escape or Close returns focus to the button. Share and copy results are announced, and if copying fails the message is selected for manual copy. Fit, crowd and confidence tones always come with an icon and words, never color alone. Reduced motion (OS or data-motion="reduce"): no FLIP, no bar growth, no spend pulse or busy blink. Forced colors: borders, use dots and bars fall back to system colors.