模型未返回最终正文（finish_reason=length），以下为推理过程摘录，可尝试提高 max_tokens 或改用多段生成。
我们需要理解任务：用户要求只输出一段可直接写入数据库的 HTML 正文片段。禁止 markdown 代码围栏、禁止解释。允许的 HTML 标签仅有 <p>、<h3>。禁止输出 <img>。不要输出 <html><body> 外壳，不要 script/style。
应用名称需要在首段 <p> 正文中自然出现：i深圳。
整体结构顺序严格：
1) 第一个元素是 <p>：依据「首段」模板改写应用介绍。首段模板原文：i深圳app是深圳市政府推出的官方政务服务平台，致力于为市民和企业提供一站式便捷服务。该应用整合了政务服务与生活服务两大板块，涵盖交警业务、社保公积金、电子证照等核心功能，并通过区块链技术保障信息安全。市民只需一个账号即可享受刷脸认证、在线办事等智能化服务，真正实现
注意：原文首段似乎不完整，以“真正实现”结尾。我们需要完成改写，作为应用介绍的第一段。要求内容包含应用名称，与原文主旨高度关联一致，逻辑清晰，更像应用商店文案。注意：首段是 <p> 而不是 h3。应该把标题也自然出现？应用名称须在首段 <p> 正文中自然出现，因此第一段可以写“i深圳是……”或“i深圳 app 是……”。
2) 然后仅按给定小节顺序输出：玩法 → 使用说明 → 优势 → 操作指南 → 特色 → 测评。但题目说随机抽 3～6 个非测评小节，并优先保留采集中有原文者；若含介绍/简介类关键词会固定排在首段后的第一个 h3；其余小节再按顺序；末位测评，全文共 4～7 个 h3。
这里给出了所有六个小节，但“随机抽 3～6 个”表示我们需要从非测评小节中抽取 3～6 个。注意：所有给出的小节都有任务描述，但末尾说“已从词池随机抽 3～6 个非测评小节并优先保留采集中有原文者；若含「介绍」「简介」类关键词会固定排在首段 <p> 之后的第一个 <h3>，其余小节再按其下列顺序；末位固定测评，全文共 4～7 个 <h3>）。” 当前列出的“小节顺序”是完整的 6 个：玩法、使用说明、优势、操作指南、特色、测评。其中非测评小结有 5 个。我们需要选择 3～6 个？ Actually “随机抽 3～6 个非测评小节”：这意味着我们要从列出的非测评小节中抽取 3 到 6 个？ 5 个，所以可抽 3、4、5 个。末位固定测评，所以 h3 数量为 4 到 6? 如果 3 个非测评+测评=4，6 个非测评+测评=7 (但非测评只有5，所以最多6)。题目说“全文共 4～7 个 <h3>”，可以 4 到 6 个 if 选3-5. But it says “随机抽 3～6 个非测评小节”， and “末位固定测评，全文共 4～7 个 <h3>” matches if 3-6 non-eval sections. Since there are only 5 non-eval sections, choose 3,4,5? We need decide how many to include.
But the subsequent “各小节任务与已代入模板” includes all 5 non-eval sections with specific counts. The overall instruction says “按下列「小节顺序」输出（已从词池随机抽 **3～6** 个非测评小节并**优先**保留采集中有原文者；若含「介绍」「简介」类关键词会固定排在首段 <p> 之后的第一个 <h3>，其余小节再按其下列顺序；末位固定测评，全文共 **4～7** 个 <h3>）”
It also says “【整体结构】顺序必须严格如下： ... 2) 然后**仅按下列「小节顺序」**输出（已从词池随机抽 **3～6** 个非测评小节...）” This “已从词池随机抽” maybe implicit: The given list is the result of random selection? Wait, they say “小节顺序：玩法 → 使用说明 → 优势 → 操作指南 → 特色 → 测评”。 That is the full order, but then says "only according to the following 'section order' output (randomly drawn 3-6?)". Hmm.
Let's parse carefully.
The initial prompt (translated):
Application name: i深圳
[Overall structure] Strict order:
1) The first element is a <p>: based on the "first paragraph" template ({head}/{text} have been substituted as first paragraph original) rewrite app intro.
2) Then **only according to the following "section order"** output (randomly selected 3 to 6 non-review sections from the word pool and **preferentially retaining those with original text in collection; if containing "introduction/简介" class keywords will be fixed after the first <p> as the first <h3>, remaining sections then by the following order; last fixed review, total 4-7 <h3>): Each section starts <h3> (title must exactly match {H3title}), non-review sections are several <p>, each point must be numbered and consistent; **the count of points in each section follows the [number of points] below** (between 3 and 6 and independent).
3) **Last section fixed as "i深圳测评"** (already at end of order): **only 1** <p>, **no** numbering, one short review, **full text no more than 200 words**.
Then "Section order: 玩法 → 使用说明 → 优势 → 操作指南 → 特色 → 测评"
Then there are task descriptions for each, with counts: 玩法 3, 使用说明 6, 优势 4, 操作指南 6, 特色 5, 测评 1.
So it seems everything is supplied. However "randomly selected 3 to 6" suggests we should choose a subset? But then "then only according to the following section order" means all six? Could contradiction. Let's inspect: It says "已从词池随机抽 3～6 个非测评小节并优先保留采集中有原文者" — this might be meta-instruction to the model: the sections listed already have been chosen? Actually “已从词池随机抽” = “has randomly selected 3-6 non-review sections from the word pool”. But they list all six section order, including review. So maybe the system has randomly selected a number of sections, but provided all? The instruction says “然后**仅按下列「小节顺序」**输出（已从词池随机抽 **3～6** 个非测评小节...）” Means we should output only according to the following order, but the number of non-review sections should be between 3 and 6. Since there are 5 non-review sections, perhaps we can include all 5? That is within 3-6. But then why say random? It might be from a larger word pool; here the actual selected sections are the five (plus eval) or a subset? Need infer.
Given each non-eval section has detailed generated template with fixed counts. If we include all five, total h3 = 6, within 4-7. If we include fewer, the omitted sections still have details but we ignore. Which is more likely? User says "按下列「小节顺序」输出" and then lists all sections. Then later provides all section sub-tasks. So likely they expect all non-review sections (5) plus eval = 6 h3. But they also emphasize randomly selected 3-6; maybe due to a template that normally randomizes but for this particular task they provide all? Let's examine: The order includes “玩法 → 使用说明 → 优势 → 操作指南 → 特色 → 测评”. Could be the exact selected sections? They say “已从词池随机抽 3～6 个非测评小节并优先保留采集中有原文者” — the phrase "已从词池随机抽" suggests the template engine has already chosen. “若含「介绍」「简介」类关键词会固定排在首段 <p> 之后的第一个 <h3>，其余小节再按其下列顺序；末位固定测评，全文共 4～7 个 <h3>” There is no "介绍" or "简介" section among the listed run sections, so no special. So the selected non-review sections likely are: 玩法, 使用说明, 优势, 操作指南, 特色? That's 5, all non-review. They might all be selected. The phrase "and prefer retaining those with original text" — none have original except first paragraph? Actually each section says "本小节无对应采集原文，按「原创H3-普通」撰写" — so no original text beyond first paragraph; thus all sections have no original, so preferential retention doesn't matter. The random selection likely happens before our prompt and yields 5 non-review sections? Wait "已从词池随机抽 3～6 个非测评小节" - perhaps the user prompt is automatically generated by a system that picks sections from a pool, and the "小节顺序" line lists the sections that were picked, in the order they should appear. Here it lists 5 non-review sections, but “3～6” means there are 5, okay. It says "仅按下列「小节顺序」输出" — i.e., only output the given sections in this order. Since all are given, include all. The phrase "末位固定测评" means one review at end. So total 6 h3. That seems plausible.
But wait: The "小节顺序" includes 6 elements, but only 5 non-review. The random selection is from non-review pool, so the selected non-review sections are the first five entries, followed by review. So output 6 h3. Good.
However there is mention "全文共 4～7 个 <h3>". 6 is within. Good. So I'll output 6 h3.
Now constraints:
- Allowed HTML tags: only <p> and <h3>.
- No markdown code fences, no explanation.
- The first element is a <p> that rewrites app intro.
- Each h3 exactly: "i深圳玩法", "i深圳使用说明", "i深圳优势", "i深圳操作指南", "i深圳特色", "i深圳测评".
- For each non-eval section: several <p>, each point numbered with "数字+全角句号．" at line start, each <p> is one point. Need ensure each point is a separate line? The requirement says “每条单独成行” and “每 <p> 一条要点”. So each <p> contains one point, text begins immediately after <p>, with “1．”, etc. The numbering should be continuous across entire full text? “全篇统一为同一种序号格式” but not necessarily continuous across sections? It says “用序号分条：每条单独成行，行首为「数字+全角句号．」（示例 1．2．3．），全篇保持同一格式”。 It doesn't say continuous numbering across sections; likely each section numbers restart from 1. It says “各小节独立” for counts; also “每节要点条数以该小节下方【条数】为准（在 3～6 条之间且各节独立）”. So each section restarts at 1. We'll do that.
- The last section "i深圳测评" has only one <p>, no numbering, a short review ≤200 characters. Need total full text ≤200? It says "全文不超过 200 字（含标点）" under测评? Let's read: “**仅 1 个** <p>，**不要**序号与分条，**一段**短评，**全文不超过 200 字**（含标点）。” Wait "全文不超过 200 字" likely refers to the review paragraph? Actually wording: "最后一个小节固定为「i深圳测评」（已在下列顺序末位）：**仅 1 个** <p>，**不要**序号与分条，**一段**短评，**全文不超过 200 字**（含标点）。" Hmm if it says full text no more than 200 characters, impossible given all sections. So it must refer to the 测评 paragraph. It says "无独立测评段，请根据首段与其它小节摘要，内化下列「原创H3-测评」模板，只输出 **一个** <p>、无序号、≤200 字。" So indeed the review paragraph ≤200. We'll keep it under 200.
- No <img>. No script/style.
Now content generation.
First paragraph: Must be a <p> containing app name "i深圳". Original first paragraph is slightly incomplete. Need rewrite. Keep related to official government service platform, integrated government services and life services, includes traffic police, social security/housing fund, e-certificates, blockchain security, one account, face recognition, online handling, etc. We should write a coherent intro, perhaps: "i深圳是深圳市官方推出的政务服务平台，整合政务服务与生活服务两大板块，提供交警业务、社保公积金、电子证照等核心功能，并通过区块链技术保障信息安全。市民只需一个账号即可刷脸认证、在线办事，真正实现“掌上办事”的便捷体验。" Need ensure “i深圳” appears naturally. Since first section doesn't have h3 title "介绍", we don't need a h3 for intro. First element p.
Need avoid too much? It can be one or more? The first element is a single <p>. It can be a paragraph.
Now non-eval sections.
We need generate based on first paragraph and the specific section themes. Each point must be "每条一句、每条约50字内". Need keep each point short, with no more than maybe 50 Chinese chars. Also must use standard numbering "1．2．3．" (full-width dot). In each section, the number starts at 1.
Let's outline:
Section 1: <h3>i深圳玩法</h3>
Focus: core mechanisms, operation, growth/loop. For a government service app, "玩法" means how to use core features. Need 3 points, each ~50 chars.
Need avoid inaccurate details? Based on first paragraph and general knowledge? Actually "无对应采集原文，按「原创H3-普通」撰写" so we can create plausible generic features. Since it's an official app, we can mention:
1．下载注册后完成实名认证，即可解锁交警、社保公积金等在线服务。
2．通过刷脸认证绑定个人账号，业务办理从提交到结果全程线上流转。
3．在首页分类入口进入电子证照或生活缴费模块，按指引完成操作即可。
Need "玩法" for app: uses one account, face recognition, online processing. That's fine.
Section 2: <h3>i深圳使用说明</h3>
Focus: installation/permissions, first use, settings/notices. 6 points. Need each point a single sentence, ~50 chars. Need be practical but not too detailed. Must be plausible. E.g.
1．在官方应用商店搜索“i深圳”下载并安装，安装时允许必要权限。
2．首次打开用手机号注册，同时完成人脸识别实名认证。
3．在“我的”页面绑定个人社保、公积金等账户信息。
4．使用服务时根据提示开启定位、相机或通知权限。
5．电子证照需在联网状态下刷新，并设置手势或指纹二次验证。
6．退出登录或换绑手机号前，先在企业/个人中心完成账号解绑。
Need ensure not overcomplicate. Let's craft carefully.
Section 3: <h3>i深圳优势</h3>
Focus: relative advantages in gameplay, interface, system, optimization. 4 points, each ~50 chars.
Possible:
1．全市政务事项一站集成，减少多个部门App间的来回切换。
2．采用区块链技术保护用户数据，信息安全感优于普通便民应用。
3．刷脸认证替代反复提交材料，办事效率明显提升。
4．界面分类清晰、操作引导完整，新手也能快速找到入口。
Need mention relative to similar apps. Good.
Section 4: <h3>i深圳操作指南</h3>
Focus: specific steps and UI operations, clear. 6 points, each ~50 chars.
Need maybe step-by-step:
1．打开App后先完成登录，在首页上方搜索框输入需要办理的业务名称。
2．点击“办事”频道，根据个人/企业类型选择对应服务分类。
3．进入业务详情页后，按提示上传材料并勾选承诺事项。
4．如需使用电子证照，在“我的证照”中完成亮码或下载。
5．提交申请后，在“我的-办事进度”中跟踪审核状态。
6．若业务需线下窗口办理，可通过App预约取号并按时到场。
This is useful.
Section 5: <h3>i深圳特色</h3>
Focus: unique differences and exclusive experience. 5 points, each ~50 chars.
Possible:
1．“一站式”整合政务服务与生活缴费，覆盖市民日常高频场景。
2．电子证照体系让身份证、驾驶证等关键证件随身携带。
3．区块链存证机制为每次申报和审核流程留下可信记录。
4．主动推送社保公积金变动提醒，重要政策信息不遗漏。
5．针对企业开办、营商审批提供专项通道，同步服务企业与个人。
Need ensure not overlap too much with others? Some overlap okay, but try to vary.
Section 6: <h3>i深圳测评</h3>
Only one <p>, no numbering, ≤200 chars. Need write a short objective review, maybe 2-3 sentences, with personification. Need incorporate highlights and recommendation. For example:
"i深圳把政务大厅装进了手机，从刷脸登录到在线办事都给人踏实感。区块键技术与电子证照让它比一般便民应用更让人放心，操作指引也足够友好。对深圳市民和企业而言，这是一只值得长期留存的“掌上办事窗口”。" Need count ≤200. It is.
Need ensure all h3 titles exactly as given. Use "i深圳" not "i深圳App" in titles. Good.
Now need check formatting: output must be directly insertable HTML fragment. So we can just output:
<p>i深圳是深圳市政府推出的官方政务服务平台，将政务服务与生活服务整合于一个账号之下，聚合交警业务、社保公积金、电子证照等高频功能，并借助区块链技术守护信息安全。市民刷脸认证后即可在线办事，真正实现“一屏通办”的便捷体验。</p>
<h3>i深圳玩法</h3>
<p>1．下载注册并完成实名认证，即可解锁交警、社保公积金等核心服务。</p>
<p>2．刷脸登录后，个人业务从申请到结果反馈均可在线上完成。</p>
<p>3．通过首页分类入口进入电子证照或生活服务，按指引逐步操作即可。</p>
<h3>i深圳使用说明</h3>
... etc.
Need ensure each point is its own <p>. No additional paragraphs. Good.
Now let's refine each point to be within 50 chars and one sentence. Need be mindful of "全篇保持同一格式" for numbering. We'll use "1．" etc. Not "1.".
Let's count characters? Not necessary for non-eval. But keep ~50. For last paragraph ≤200. Let's draft all.
First paragraph maybe:
"i深圳是深圳市政府推出的官方政务服务平台，将政务服务与生活服务整合进一个账号，聚合交警、社保公积金、电子证照等高频功能，并通过区块链技术保障信息安全。市民刷脸认证后即可在线办事，真正实现“一屏通办”的智慧体验。"
This is about 85 chars. Good.
Now section 1 "玩法": Need 3 points. Let's craft:
1．下载注册并完成实名认证，即可解锁交警、社保公积金等核心服务。
2．通过刷脸认证登录个人账号，业务申请、材料提交与进度查询全程线上流转。
3．在首页“办事”分类进入对应服务，按步骤填写信息即可完成一次线上办理。
Each perhaps <50? Let's estimate: first 30, second 35, third 30. good.
Section 2 "使用说明": 6 points.
Need "安装与权限、首次上手、常见设置与注意事项". Let's craft:
1．在官方应用商店搜索“i深圳”下载，首次安装按系统提示开启必要权限。
2．打开App后用手机号注册，并配合人脸识别完成实名认证。
3．在“我的”栏目中绑定社保、公积金等常用账户，便于后续快速查询。
4．使用定位、扫一扫或相机功能时，根据页面提示授权对应权限。
5．可在设置中开启消息通知，及时获取事项进度与政策提醒。
6．更换设备或退出登录前，先在设置中解绑旧设备并清理缓存信息。
This is 6, practical. Need ensure no weird "设置中" may exist? It's fine.
Section 3 "优势": 4 points.
Let's craft:
1．政务事项一站汇聚，免去在多个部门App之间反复切换的麻烦。
2．区块键技术为信息加密存证，数据防护比普通便民应用更可靠。
3．刷脸认证与电子证照结合，提交材料少、审核效率高。
4．操作界面和办事指引经过统一设计，首次使用也能快速定位服务。
Need "区块键" typo? Should be "区块链". Good.
Section 4 "操作指南": 6 points.
Need specific steps:
1．登录后在首页顶部搜索框输入业务名称，可直达对应办理入口。
2．点开“办事”频道，按“个人办事”或“企业办事”选择服务分类。
3．进入业务详情页，先阅读办理条件，再按提示上传所需材料。
4．使用电子证照时，在“我的证照”中选择需亮码的证件并出示。
5．提交申请后进入“我的-办事进度”，即可跟踪当前审核环节。
6．需要线下办理的服务，可在App上预约取号并按约定时间到场。
Good.
Section 5 "特色": 5 points.
Need "相对同类作品最突出的差异点与独有体验":
1．政务服务与生活缴费双板块合一，覆盖市民日常高频场景。
2．电子证照体系将身份证、驾驶证等常用证件装进手机。
3．区块链存证为每次申报与审批流转留下不可篡改记录。
4．主动推送社保公积金变动和政策消息，减少信息遗漏焦虑。
5．为企业开办与经营事项提供专属通道，个人和企业都能用。
Need maybe "单独为港澳/外国人"? No.
Section 6 "测评": one p, no number, ≤200. Need maybe a bit personal.
Let's craft:
"体验i深圳时，最直观的感受是把政务大厅搬进了手机。刷脸登录、在线办事和电子证照一气呵成，区块链背书也让隐私安全更有底气。界面清晰、流程顺畅，既适合日常查询，也能高效完成申报，是深圳人值得信赖的掌上服务助手。"
Count approx: "体验i深圳时，最直观的感受是把政务大厅搬进了手机。" 25? Let's count manually? Not necessary. Probably under 200.
Need ensure "全文不超过200字" for测评. It is ~100.
Now consider if any section should be excluded? Let's revisit "随机抽 3～6 个非测评小节" — If include all 5, it's within 3-6. But "仅按下列「小节顺序」输出" says all. Good.
Need avoid extra tags like <br>. Each <p> is its own line. No blank lines? Not an issue. We can output just elements.
Need make sure there is no markdown code fences. We'll output raw HTML.
Potential issue: The first paragraph should be "一个 <p>：依据下方「首段」模板（{head}/{text} 均已代入为首段原文）改写应用介绍。" It says "第一个元素是一个 <p>" not h3. So no "i深圳" as h3 before first p. The app name must appear in first p. Yes.
Need check "禁止输出 <img>（应用截图由系统在生成后按小节顺序自动插入，模型无需处理图片）" No img.
Need check "允许的 HTML 标签仅有：<p>、<h3>" So only p and h3. We comply.
Need check "不要输出 <html><body> 等外壳" comply.
Now let's think if "每条单独成行，行首为「数字+全角句号．」" with <p> tags, "行首" might be ambiguous: Inside the HTML, each <p> begins on a new line and its text begins with "1．". We'll ensure that.
Need maybe each h3 followed directly by <p> and no extra blank lines? Fine.
Need use "全角句号．" after digit, e.g. "1．" The Chinese full-width dot is "．" (U+FF0E). We'll use that.
Need "全篇统一为同一种序号格式" — we use "1．" etc.
Need "每节要点条数按【条数】": 3 for玩法, 6 for使用说明, 4 for优势, 6 for操作指南, 5 for特色. Good.
Now let's ensure all points are independent and not too overlapping across sections. There may be some repetition: "刷脸认证" appears in first paragraph, 玩法2, 优势3, 测评. That's okay but maybe repetitive. We can vary. "电子证照" appears in 玩法3, 优势3, 特色2, 操作指南4, 测评. This might be too repetitive. But it's a core feature; okay. Need perhaps reduce repetitions by focusing different aspects.
Let's refine:
First paragraph: mention core aspects: official platform, one account, services, blockchain, face recognition, one-screen.
玩法: "核心机制、操作方式、成长/对局节奏与主要游戏循环" - For app, maybe "通关" metaphor? But better not too gamey? It says game/software, but "玩法" for software can be "怎么玩". Could use "玩法" as "use". We can write:
1．完成手机号注册与人脸识别实名认证后，即可开启一站式办事体验。
2．以个人账号为中心，在“办事”“生活”两大频道间切换处理事务。
3．每项业务按“信息填写—材料上传—进度跟踪”的流程完成闭环。
This is more like core loop.
Use "使用说明" should focus on installation and settings. Good.
"优势" should focus on comparison. Need avoid "界面清晰" in advantage vs Operation guide; okay.
"操作指南" should be step-by-step. Good.
"特色" should focus on unique features. Good.
Let's refine all text to be polished, proper Chinese, and no factual claims too strong. All based on general knowledge of i深圳? Actually i深圳 is real app? Yes, Shenzhen government service. Features include 统一实名认证, 电子证照, etc. It's okay.
Let's craft final output.
But before final, let's consider whether "首段" must contain "i深圳" exactly perhaps with "App" or "app". It says application name: i深圳. The first paragraph "应用名称（须在首段 <p> 正文中自然出现）：i深圳" So include "i深圳" exactly. We'll start "i深圳是深圳市官方推出的..." Good.
Need not include "app" after i深圳? It says "i深圳app" in original; but name is "i深圳". We'll write "i深圳是..." That's natural. Could also write "i深圳 App" but if name is "i深圳", maybe better exact. We'll use "i深圳" exactly.
Now think about possible issue: "h3" titles should be exactly "i深圳玩法", "i深圳使用说明", etc. We'll use no attributes.
Now maybe include "i深圳测评" as last h3. Good.
Now let's write final content in raw HTML. Ensure no leading/trailing text. We'll output as:
<p>i深圳是深圳市政府推出的官方政务服务平台，将政务服务与生活服务整合到同一个账号体系，汇聚交警、社保公积金、电子证照等高频功能，并用区块链技术保障信息安全。市民刷脸认证后即可在线办事，真正实现“一屏通办”的智能体验。</p>
<h3>i深圳玩法</h3>
<p>1．完成手机号注册和人脸识别实名认证，解锁一站式线上办事服务。</p>
<p>2．以个人账号为中心，在“办事”和“生活”两大频道间切换处理日常事务。</p>
<p>3．每项业务按信息填写、材料上传、进度跟踪的流程形成完整闭环。</p>
<h3>i深圳使用说明</h3>
<p>1．在官方应用商店搜索“i深圳”下载安装，并按系统提示开启必要权限。</p>
<p>2．首次打开使用手机号注册，配合人脸识别完成实名认证。</p>
<p>3．在“我的”页面绑定社保、公积金等常用账户，便于快速查询和办理。</p>
<p>4．使用定位、扫一扫或相机功能时，根据页面指引授权相应权限。</p>
<p>5．建议在设置中开启通知提醒，及时获取办事进度和政策动态。</p>
<p>6．更换设备或退出登录前，先解绑旧设备并清理本地缓存信息。</p>
<h3>i深圳优势</h3>
<p>1．政务事项一站汇聚，免去在多个部门App之间反复切换的麻烦。</p>
<p>2．区块链技术为信息加密存证，数据防护比普通便民应用更可靠。</p>
<p>3．刷脸认证与电子证照结合，减少材料提交，提升办事效率。</p>
<p>4．统一设计的操作入口和引导，让不同年龄用户都能快速上手。</p>
<h3>i深圳操作指南</h3>
<p>1．登录后在首页顶部搜索框输入业务名称，系统会直达对应办理入口。</p>
<p>2．点开“办事”频道，根据“个人办事”或“企业办事”选择服务分类。</p>
<p>3．进入业务详情页后先阅读办理条件，再按提示上传所需材料。</p>
<p>4．使用电子证照时，在“我的证照”中选择相应证件并完成亮码。</p>
<p>5．提交申请后，可在“我的-办事进度”中实时跟踪审核状态。</p>
<p>6．如需线下办理，可在App上预约取号并按约定时间到达窗口。</p>
<h3>i深圳特色</h3>
<p>1．政务服务与生活缴费双板块合一，覆盖市民日常高频场景。</p>
<p>2．电子证照体系把身份证、驾驶证等常用证件装进手机随取随用。</p>
<p>3．区块链存证为每次申报和审批流转留下不可篡改的记录。</p>
<p>4．主动推送社保公积金变动及重要政策，减少错过关键信息。</p>
<p>5．为企业开设和经营事项打造专属通道，个人与企业都能受惠。</p>
<h3>i深圳测评</h3>
<p>体验i深圳时，最直观的感受是把政务大厅搬进了手机。刷脸登录、在线办事和电子证照一气呵成，区块链背书也让隐私安全更有底气。界面清晰、流程顺畅，既适合日常查询，也能高效完成申报，是深圳人值得信赖的掌上服务助手。</p>
Need check "全文不超过 200 字" for测评: It's around 100. Good.
Now let's evaluate if "首段" needs to be "对原文进行原创改写，内容必须包含游戏/软件名称，与原文的表达主旨保持高度关联与一致性。可以根据原文的内容进行内容重组，使其逻辑清晰、层次分明，更偏向于一个应用商店的介绍文案。" Our first paragraph is original rewrite, includes name, high relevance. Good.
Potential issue: "i深圳app" original says "app"; our text "i深圳是..." okay.
Potential issue: The first section h3 after first p is "i深圳玩法" because no intro section. Good.
Now let's consider if "随机抽 3～6 个非测评小节" might mean we should not include all 5 if random selection chosen some subset. But the output we give includes all 5, which is a valid count in 3-6. However the instruction "已从词池随机抽 3～6 个非测评小节" could be interpreted as "you should randomly pick 3-6" but that would conflict with "only according to the following section order". Since we are a model, we shouldn't randomly vary; we should follow explicitly given order. The supplied order includes all sections, so include them. Good.
But let's not ignore the phrase "**优先**保留采集中有原文者；若含「介绍」「简介」类关键词会固定排在首段 <p> 之后的第一个 <h3>，其余小节再按其下列顺序" There is no "介绍" section, so no special. So the first non-review section is "玩法" as per order. Good.
Let's verify "末位固定测评，全文共 4～7 个 <h3>" - Our total h3 = 6. Good.
Now need output only one fragment. We'll ensure no "```html" etc.
Let's also consider whether to include a blank line between elements? Not necessary. But "可直接写入数据库的 HTML 正文片段" means raw text okay. We can include newlines for readability. Fine.
Potential issue: In "操作指南" Point 3: "进入业务详情页后先阅读办理条件，再按提示上传所需材料。" This is good. Point 6: "如需线下办理，可在App上预约取号并按约定时间到达窗口。" "App" might not be "i深圳" but okay. Should we avoid English "App"? Allowed. But "App" is not an HTML tag. Fine.
Need ensure every non-eval point is a <p> and no other tags like <ul>. Yes.
Now let's think if "每条约50字内" for all non-eval points. Let's count some to ensure they are roughly under 50. Some are around 30. Good.
Let's double-check "全角句号．" after digit. It may be "1．" but the dot is full width. We use a full width dot. Good.
Potential issue: The Chinese punctuation "．" after the number should be followed immediately by content, no space. We'll do.
Now let's refine content to be concise and natural, avoid over-claiming "一键". The original says "真正实现" (likely "最多跑一次" or "一次都不跑"). We wrote "一屏通办" maybe invented. But it's a reasonable app store copy. Could be okay. To align with original, "真正实现" maybe "真正实现办事少跑腿" more natural. Let's adjust first paragraph:
"i深圳是深圳市政府推出的官方政务服务平台，将政务服务与生活服务整合到同一个账号，聚合交警业务、社保公积金、电子证照等高频功能，并通过区块链技术保障信息安全。市民刷脸认证后即可在线办事，真正实现政务事项一站通办、少跑快办。"
This is good. But "i深圳是深圳市政府推出的官方政务服务平台" quite close to original; need "原创改写" maybe not too close? It's okay but perhaps "由深圳市官方推出" etc. Let's make more original:
"由深圳市政府官方打造的i深圳，是一站式政务与生活服务平台。它把交警业务、社保公积金、电子证照等高频事项汇聚在一个账号下，并用区块链守护信息安全，市民刷脸认证后即可在线办事，享受到更高效、更省心的城市服务。"
This includes name in "i深圳" not at beginning? It says "应用名称须在首段正文中自然出现" so okay. The first element is a <p>. This is a rewrite. Maybe more app store style. Let's use this.
Need "原文首段：i深圳app是..." Perhaps "i深圳app" original. Our first sentence "由深圳市政府官方打造的i深圳" natural. Good.
Let's update first paragraph:
<p>由深圳市政府官方打造的i深圳，是一站式政务与生活服务平台。它把交警业务、社保公积金、电子证照等高频事项汇聚在一个账号下，并用区块链守护信息安全，市民刷脸认证后即可在线办事，享受到更高效、更省心的城市服务。</p>
This contains "i深圳" naturally. Good.
Need perhaps "首段原文" says "真正实现" incomplete; we completed it. Fine.
Now check "应用名称（须在首段 <p> 正文中自然出现）：i深圳" done.
Now let's think if all section contents are based on first paragraph and not invented beyond? The "操作指南" mentions search box, "办事" channel, etc. These are generic but plausible. The instruction says "可参考信息：i深圳app是..." and "本小节无对应采集原文，按原创H3-普通撰写" so allowed.
Now perhaps "测评" paragraph: "区块链背书" might sound too technical but good.
Let's count "测评" paragraph characters to ensure ≤200:
体验i深圳时，最直观的感受是把政务大厅搬进了手机。刷脸登录、在线办事和电子证照一气呵成，区块链背书也让隐私安全更有底气。界面清晰、流程顺畅，既适合日常查询，也能高效完成 
