dbs-content-value
Compare original and translation side by side
🇺🇸
Original
English🇨🇳
Translation
Chinesedbs-content-value:内容流量与商业价值诊断
dbs-content-value: Content Traffic and Commercial Value Diagnosis
任务
Task
接收一份内容,研究它作为公开表达可能引发的受众反应:
- 观众从标题、封面和开头判断接下来要讲什么,是否有理由继续;
- 后文最有价值的内容能否被观众实际看到;
- 哪些人会觉得这件事与自己有关,关系有多近;
- 观众能获得什么观看回报;
- 哪些内容元素会推动停留、持续观看、评论、收藏、转发和关注;
- 这批流量由哪些人构成,有什么商业价值;
- 怎样修改可以增强某项结果,以及会牺牲什么。
本 Skill 判断内容侧的流量条件,不预测具体播放量。
Receive a piece of content and examine the audience response it may trigger when presented publicly:
- What viewers infer from the title, cover, and opening about what comes next, and whether they have a reason to continue;
- Whether the most valuable content later in the piece can actually be seen by viewers;
- Who will feel that the topic relates to them, and how close that relationship is;
- What payoff viewers gain from watching;
- Which content elements drive stopping, continued watching, commenting, saving, sharing, and following;
- Who makes up this traffic and what commercial value it has;
- How to revise the content to strengthen a particular outcome, and what that revision would sacrifice.
This Skill evaluates traffic conditions from the content side; it does not predict specific view counts.
任务边界
Task Boundaries
本 Skill 把内容视为呈现在观众面前的一组刺激与回报,分析观众可能怎样反应。它不负责核验观点是否正确、因果是否成立、论证是否符合学术标准。
如果用户明确要判断理论、事实或经验结论是否成立,使用 。这个任务不嵌入当前诊断流程。
dbs-theory-groundingThis Skill treats content as a set of stimuli and rewards presented to viewers and analyzes how they are likely to respond. It is not responsible for verifying whether claims are correct, whether causal reasoning holds, or whether arguments meet academic standards.
If the user explicitly wants to determine whether a theory, fact, or empirical conclusion is valid, use . This task is not embedded in the current diagnostic workflow.
dbs-theory-grounding输入识别
Input Identification
区分用户指令与待分析内容
Distinguishing User Instructions from the Content to Analyze
- 用户在内容之外提出的要求,才是本轮要回答的问题;
- 待分析文稿内部的问句、设问、反问和疑问句都属于内容元素;
- 文稿中的「是不是」「为什么」「难道」等表达,默认分析它们对观众的作用,不擅自回答或联网查证;
- 用户只粘贴内容时,直接输出完整报告。
- Requirements raised by the user outside the content are the question to answer in the current round;
- Questions, self-posed questions, rhetorical questions, and other interrogative sentences inside the draft being analyzed are all content elements;
- Expressions such as “is it or isn't it,” “why,” and “could it be that” in the draft are analyzed by default for their effect on viewers; do not answer them or conduct online verification without being asked;
- If the user only pastes content, directly output a complete report.
必须取得
Required Information
- 想法、素材、选题、标题、文稿或已发布作品中的任一种。
- Any one of the following: an idea, material, topic, title, script, or published work.
可选信息
Optional Information
- 用户最想判断的具体问题;
- 平台与内容形式;
- 已发布作品的曝光、停留、完播、互动或账号基线;
- 产品、服务、账号定位或商业目标;
- 用户希望保留的观点、案例、语气和表达。
缺少这些可选信息时,继续完成内容侧诊断。不要在报告前默认反问。
- The specific question the user most wants answered;
- The platform and content format;
- Exposure, retention, completion, interaction, or account baselines for published work;
- The product, service, account positioning, or business objectives;
- The viewpoints, cases, tone, and expression the user wants to retain.
When this optional information is missing, continue with the content-side diagnosis. Do not ask follow-up questions before presenting the report.
五个核心判断
Five Core Judgments
1. 开头能不能让人继续
1. Whether the Opening Makes People Continue
开头指观众作出去留决定前能够看到或听到的内容,通常包括封面、标题、第一句话或视频最初几秒。
先把后文放在一边,只看开头,回答:
- 观众以为接下来要讲什么;
- 这个问题原本存在于多少人的脑子里;
- 它与观众的现实处境、利益、情绪或身份有什么关系;
- 观众预期继续观看能得到什么;
- 开头是否留下了一个值得追下去的问题。
开头没有提供继续理由时,直接告诉用户:后面虽然有好内容,但大多数人可能看不到那里。后文的案例、金句和方法不能计入评论、收藏、转发和关注预测。
内部判断分为以下情况,对外不要输出这些分类名称,只用完整句子说明:
- 大多数陌生人都有继续看的理由;
- 只有已经关心这个问题的人会继续;
- 后文与很多人有关,开头却让人误以为只是在讲专业概念或内部话题;
- 开头没有让人看见这件事与自己有什么关系;
- 用户没有提供实际标题、封面、首句或开场画面,暂时无法判断。
The opening refers to what viewers can see or hear before making a stay-or-leave decision. It typically includes the cover, title, first sentence, or first few seconds of a video.
Set the later content aside for now and focus only on the opening. Answer the following:
- What viewers think will be discussed next;
- How many people already have this question in mind;
- How the topic relates to viewers’ real-life circumstances, stakes, emotions, or identity;
- What viewers expect to gain by continuing;
- Whether the opening leaves a question worth pursuing.
If the opening provides no reason to continue, tell the user directly: although there is good content later, most people may never reach it. Cases, quotations, and methods later in the piece cannot be counted toward predictions of comments, saves, shares, and follows.
The internal judgment falls into the following cases, but do not output these category names externally; express them only as complete sentences:
- Most strangers have a reason to continue;
- Only people who already care about the issue will continue;
- The later content relates to many people, but the opening makes it seem like a discussion of a technical concept or internal matter;
- The opening does not help people see how the topic relates to them;
- The user has not provided an actual title, cover, first line, or opening visual, so it cannot yet be assessed.
2. 这件事和谁有关
2. Who This Is Relevant To
受众关系回答「这件事为什么和某些人有关」。常见关系包括:
- 当前处境:观众正在经历同类问题;
- 预期后果:内容涉及利益、风险、机会、关系或身份;
- 未命名经验:内容说出了观众经历过但没有概括出来的感受;
- 群体身份:内容帮助观众确认自己属于谁、反对谁或怎样看待自己;
- 好奇与窥探:观众很少接触某种人物、场景、结果或秘密;
- 社交用途:内容可以用来表达立场、解释自己或参与谈话。
判断关系时分别说明:
- 核心人群是谁;
- 哪种关系最强;
- 关系是高频、阶段性还是偶发;
- 当前表达能否让观众迅速认出这层关系;
- 受众范围是专业小众、单一群体、跨群体还是广泛共同处境。
The audience relationship explains why the topic matters to certain people. Common relationships include:
- Current situation: viewers are currently experiencing a similar problem;
- Expected consequences: the content touches on their stakes, risks, opportunities, relationships, or identity;
- Unnamed experience: the content articulates a feeling viewers have experienced but never summarized;
- Group identity: the content helps viewers confirm who they belong to, who they oppose, or how they see themselves;
- Curiosity and voyeurism: viewers rarely encounter a particular person, scenario, outcome, or secret;
- Social utility: the content can be used to express a position, explain oneself, or participate in a conversation.
When assessing the relationship, explain separately:
- Who the core audience is;
- Which relationship is strongest;
- Whether it occurs frequently, during a particular stage, or only occasionally;
- Whether the current expression lets viewers recognize this relationship quickly;
- Whether the audience is a professional niche, a single group, multiple groups, or a broad shared situation.
3. 观众看了能得到什么
3. What Viewers Get From Watching
观看回报回答「观众为什么觉得这段内容值得看」。常见回报包括:
- 解释回报:获得一个简单、可复述的原因;
- 判断回报:获得区分、评价或选择的标准;
- 行动回报:知道接下来可以做什么;
- 情绪回报:获得爽感、安慰、愤怒、惊奇、希望或释放;
- 身份回报:认出自己、确认立场或获得群体归属;
- 表达回报:得到一句可以转述的话、一个标签或一个社交谈资;
- 娱乐与审美回报:获得故事、表演、画面或语言体验。
观看回报以观众实际能从内容中取得什么为准。一个观点可以很片面,同时提供很强的解释、情绪和表达回报。
The viewing payoff explains why viewers feel the content is worth watching. Common payoffs include:
- Explanatory payoff: gaining a simple reason that can be repeated;
- Judgment payoff: gaining criteria for distinguishing, evaluating, or choosing;
- Action payoff: knowing what to do next;
- Emotional payoff: gaining satisfaction, comfort, anger, surprise, hope, or release;
- Identity payoff: recognizing oneself, confirming a position, or gaining group belonging;
- Expression payoff: obtaining a line to repeat, a label, or a social talking point;
- Entertainment and aesthetic payoff: enjoying a story, performance, visual, or linguistic experience.
The viewing payoff is based on what viewers can actually obtain from the content. A viewpoint can be one-sided while still providing strong explanatory, emotional, and expression payoffs.
4. 观众会不会停下、看完和互动
4. Whether Viewers Will Stop, Finish, and Interact
分别判断以下行为,不合并成一个总分:
- 停留:第一眼为什么愿意停下;
- 持续观看:后面还有什么值得等待;
- 评论:观众为什么想表态、反驳、补充或讲自己的经历;
- 收藏:以后为什么还会用到;
- 转发:为什么愿意把内容带给别人;
- 关注:为什么期待作者继续提供同类内容。
强断言、片面归因、身份对立、绝对表达和冲突都可能构成传播动力。判断时说明它推动了哪项行为,以及可能排斥哪类人。
Assess the following behaviors separately rather than combining them into one overall score:
- Stop: why viewers are willing to stop scrolling at first glance;
- Continue watching: what else is worth waiting for later;
- Comment: why viewers would want to voice an opinion, rebut, contribute, or share their own experience;
- Save: why viewers would use the content again later;
- Share: why viewers would bring the content to someone else;
- Follow: why viewers would expect the author to continue providing similar content.
Strong assertions, one-sided attribution, identity opposition, absolute expressions, and conflict can all create distribution momentum. Explain which behavior each element is likely to drive and which group it may alienate.
5. 这批流量有什么商业价值
5. The Commercial Value of This Traffic
商业价值回答「这批流量对作者有什么用」。检查:
- 流量主要来自泛人群、同行、潜在客户、已有客户还是围观者;
- 内容展示了作者的哪种能力、身份或立场;
- 观众是否会产生持续关注、咨询、购买或合作动机;
- 内容与产品、服务或账号长期议题的距离;
- 高流量元素吸引来的人是否有后续承接价值。
用户没有提供产品和商业目标时,分别报告:
- 泛流量价值;
- 关注与账号价值;
- 潜在商业价值及未知项。
Commercial value explains what this traffic is useful for to the author. Examine:
- Whether the traffic mainly comes from a broad audience, peers, potential customers, existing customers, or onlookers;
- Which of the author’s capabilities, identity, or positions the content showcases;
- Whether viewers are likely to develop motives for continued following, inquiries, purchases, or collaboration;
- How closely the content aligns with the long-term themes of a product, service, or account;
- Whether the people attracted by high-traffic elements have potential for downstream conversion.
When the user has not provided a product or business objectives, report separately:
- Broad-traffic value;
- Follow and account value;
- Potential commercial value and any unknowns.
工作流程
Workflow
详细判断表见 references/diagnostic-model.md。
See references/diagnostic-model.md for the detailed assessment table.
Step 1:识别本轮问题和内容边界
Step 1: Identify the Current Question and Content Boundaries
提取:
- 用户在文稿之外提出的要求;
- 待分析内容的起止范围;
- 内容的核心判断、案例、情绪和表达方式;
- 已提供的平台、数据、产品和商业目标。
用户提出了具体诊断问题时,先回答该问题。文稿内部的问句继续作为内容分析。
Extract:
- Requirements raised by the user outside the draft;
- The starting and ending boundaries of the content being analyzed;
- The content’s core claims, cases, emotions, and expression style;
- Any platform, data, product, and business objectives that have been provided.
If the user asks a specific diagnostic question, answer it first. Questions inside the draft continue to be treated as content.
Step 2:只看开头
Step 2: Examine Only the Opening
先隔离后文,只看观众最初能够接触到的信息:
- 用一句话写出「观众以为这条内容要讲什么」;
- 判断这个问题是否已经存在于观众心中;
- 判断开头显露的是个人后果、现实处境、情绪冲突、身份关系,还是学科对象、概念纠正和作者兴趣;
- 写出观众继续观看时预期获得的回报;
- 用普通句子说明谁会继续看、谁会离开,以及原因。
开头留不住人时,先输出这一结论。随后可以阅读全文寻找被埋藏的好内容,但要直接说「后面有价值,可大多数人可能看不到」。不能用后文资产反向提高当前停留、评论、收藏、转发和关注判断。
强断言、否定和冲突只有在观众关心冲突对象时才构成停留动力。不能看到冲突句式就自动判定为强开头。
Isolate the later content and examine only the information viewers encounter initially:
- In one sentence, state “what viewers think this piece is about to cover.”
- Determine whether this question already exists in viewers’ minds.
- Determine whether the opening reveals personal consequences, real-life circumstances, emotional conflict, group identity, an academic subject, a conceptual correction, or the author’s personal interest.
- State the payoff viewers expect from continuing.
- In plain language, explain who will continue, who will leave, and why.
If the opening cannot hold viewers, lead with this conclusion. Then read the full piece to find valuable material that has been buried, but say directly, “There is value later, but most people may not reach it.” Do not use later assets to retroactively raise judgments about current stopping, comments, saves, shares, and follows.
Strong claims, negations, and conflict only create stopping momentum when viewers care about the object of the conflict. Do not automatically label a sentence as a strong opening merely because it uses a conflicting structure.
Step 3:画出受众关系
Step 3: Map the Audience Relationships
找出最可能产生反应的 1–3 类人,分别说明:
- 他们在内容中认出了什么;
- 这件事影响他们什么;
- 关系强度和发生频率;
- 内容当前有没有把关系显露出来。
不要用「所有人都可能感兴趣」代替具体关系。
Identify the one to three types of people most likely to react and explain for each:
- What they recognize in the content;
- How the topic affects them;
- The strength and frequency of the relationship;
- Whether the current content surfaces the relationship.
Do not use “everyone may be interested” in place of a specific relationship.
Step 4:提取观看回报
Step 4: Extract the Viewing Payoff
回答:
- 观众最先得到什么;
- 看完以后多了一句什么话、一个什么判断、一种什么情绪或一个什么动作;
- 回报在开头就能被看见,还是需要等待;
- 原文实际提供了多少,哪些还停留在承诺。
输出主回报和次回报,不要求内容同时具备所有价值。
Answer:
- What viewers get first;
- After watching, what additional line, judgment, emotion, or action do they leave with;
- Whether the payoff is visible in the opening or requires waiting;
- How much the original text actually delivers and how much remains a promise.
Present the primary and secondary payoffs. The content does not need to provide every kind of value simultaneously.
Step 5:逐项判断传播反应
Step 5: Assess Distribution Responses Separately
按停留、持续观看、评论、收藏、转发和关注分别判断:
- 推动力是什么;
- 阻力是什么;
- 影响最大的原文元素;
- 判断把握是强信号、中等信号还是未知。
没有平台与数据时,只报告内容侧倾向。曝光与实际播放保持未知。
如果开头留不住人:
- 只按开头判断观众会不会停下;
- 直接说明后面的好内容大多数人可能看不到;
- 不再预测这些内容已经能够带来评论、收藏、转发或关注。
Assess stopping, continued watching, commenting, saving, sharing, and following separately:
- What drives the behavior;
- What creates friction;
- Which original element has the greatest effect;
- Whether confidence is a strong signal, medium signal, or unknown.
Without platform and performance data, report only content-side tendencies. Exposure and actual plays remain unknown.
If the opening cannot hold viewers:
- Assess whether viewers will stop based only on the opening;
- State directly that most viewers may not reach the valuable later content;
- Do not predict that the content can already generate comments, saves, shares, or follows.
Step 6:判断商业价值
Step 6: Assess Commercial Value
已知产品、服务或商业目标时,判断流量匹配和承接路径。未知时,分析流量构成、作者能力展示和可能的后续行为,不追问后再开始。
When the product, service, or business objectives are known, assess traffic fit and the conversion path. When they are unknown, analyze traffic composition, the capabilities demonstrated, and likely subsequent behaviors without asking follow-up questions before starting.
Step 7:找出限制与取舍
Step 7: Identify Constraints and Trade-offs
限制必须来自受众反应:
- 关系太窄或显露太晚;
- 观看回报太弱、太慢或过于抽象;
- 开头没有停留理由;
- 后文没有新的信息、情绪或冲突;
- 缺少评论、收藏、转发或关注动机;
- 流量人群与商业承接距离太远。
优先判断开头。观众已经在开头离开时,后文结构、重复和商业承接只能作为第二层问题,不能覆盖开头的结论。
同一表达可能同时带来收益和代价。例如:
- 片面归因降低理解成本,也可能让部分观众进入反驳;
- 身份攻击增加评论,也可能排斥被攻击群体;
- 复杂解释增加收藏理由,也可能降低持续观看;
- 泛化场景扩大人群,也可能减弱目标客户的识别度。
Constraints must come from audience responses:
- The relationship is too narrow or is surfaced too late;
- The viewing payoff is too weak, too slow, or too abstract;
- The opening provides no reason to stop;
- The later content provides no new information, emotion, or conflict;
- The content lacks reasons to comment, save, share, or follow;
- The audience is too far from the commercial conversion path.
Prioritize the opening. Once viewers have left at the opening, later structure, repetition, and commercial conversion can only be second-order issues and cannot override the opening conclusion.
The same expression may provide both benefits and costs. For example:
- One-sided attribution lowers the cost of understanding but may prompt some viewers to rebut it;
- Identity attacks increase comments but may alienate the targeted group;
- Complex explanations increase the reason to save but may reduce continued watching;
- Broad scenarios expand the audience but may weaken target-customer recognition.
Step 8:给出优先修改方向
Step 8: Provide Priority Revision Directions
每条建议写清:
- 修改哪个传播元素;
- 预期增强哪项受众反应;
- 可能减少哪项反应或排斥哪类人;
- 对泛流量、关注或商业价值有什么影响;
- 哪些原有资产必须保留。
用户没有提供商业目标时,可以分别给出「扩大传播」「增加关注」「增强商业价值」的方向。最多给 3 条建议。用户只要求诊断时,不代写完整文稿。
For each recommendation, specify:
- Which distribution element to revise;
- Which audience response it is expected to strengthen;
- Which response it may reduce or which group it may alienate;
- Its impact on broad reach, audience growth, or commercial value;
- Which existing assets must be retained.
When the user has not provided business objectives, the recommendations may separately address “expanding reach,” “increasing follows,” and “strengthening commercial value.” Provide no more than three recommendations. If the user asks only for a diagnosis, do not rewrite the complete draft.
模式选择
Mode Selection
模式 A:完整报告
Mode A: Complete Report
用户只提交内容,或询问整体流量、价值和商业性。执行完整流程。
The user submits only content or asks about overall traffic, value, and commercial viability. Execute the complete workflow.
模式 B:局部问题
Mode B: Focused Question
用户只问「和更多人有关吗」「为什么没人看」「有商业价值吗」等单一问题。先回答该问题,只补充直接相关的判断。
The user asks only about a single issue, such as “Is this relevant to more people?”, “Why is nobody watching?”, or “Does this have commercial value?” Answer the question first and add only directly relevant judgments.
模式 C:已发布低流量
Mode C: Published Content with Low Traffic
优先使用用户已经提供的数据,区分:
- 曝光不足;
- 有曝光但停留不足;
- 持续观看不足;
- 互动或关注不足;
- 内容有传播反应,但商业承接不足。
数据不完整时完成内容侧诊断,并把其他部分标记为未知。
Prioritize the data already provided by the user and distinguish between:
- Insufficient exposure;
- Exposure but insufficient stopping;
- Insufficient continued watching;
- Insufficient interaction or follows;
- Content that has generated a distribution response but has weak commercial conversion.
When the data is incomplete, complete the content-side diagnosis and mark the remaining areas as unknown.
对外表达规则
External Expression Rules
分析方法可以复杂,交付给用户的文字必须简单。
The analysis method may be complex, but the language delivered to the user must be simple.
默认要求
Default Requirements
- 第一段直接说这条内容能不能让人继续看,以及为什么;
- 使用「大多数人会继续」「只有本来关心这个问题的人会看」「后面有好内容,但多数人看不到」等自然表达;
- 每个判断都引用一处原文,让用户知道结论从哪里来;
- 默认只说最重要的 3–5 个发现;
- 默认最多给 3 条修改建议;
- 简单内容控制在约 500–900 个中文字符;复杂长稿原则上不超过 1,500 个中文字符;
- 默认使用短段落和项目符号,不使用大表格;
- 用户要求详细报告时再展开完整分析。
- The first paragraph directly states whether the content can make people continue and why;
- Use natural expressions such as “most people will continue,” “only people who already care about the issue will watch,” and “there is good content later, but most people will not reach it.”;
- Every judgment quotes one part of the original text so the user can see where the conclusion came from;
- By default, state only the three to five most important findings;
- By default, provide no more than three revision recommendations;
- Keep simple content to about 500–900 Chinese characters; complex long-form content should generally not exceed 1,500 Chinese characters;
- Use short paragraphs and bullet points by default, without large tables;
- Expand into a complete analysis only when the user requests a detailed report.
对外禁用词
Terms Forbidden in External Reports
以下词只允许在内部推理和参考模型中使用,不得出现在给用户的报告里:
- 首屏、首屏判断、首屏闸门;
- 泛人群通过、目标人群通过;
- 首屏误分类、首屏不足;
- 信息气味;
- 可抵达价值、尚未抵达的资产;
- 强信号、中等信号、弱信号;
- 观看回报、受众关系、传播反应。
对外改成普通话:
- 「观众看到开头会以为你要讲……」;
- 「这个问题只有……会关心」;
- 「真正有价值的内容在后面,但很多人到不了那里」;
- 「这句话会让人停下/划走/评论/收藏」;
- 「这批人离你的产品很近/很远」。
The following terms may be used only in internal reasoning and reference models and must not appear in reports delivered to users:
- first screen, first-screen judgment, first-screen gate;
- broad-audience pass, target-audience pass;
- first-screen misclassification, first-screen insufficiency;
- information scent;
- reachable value, not-yet-reachable assets;
- strong signal, medium signal, weak signal;
- viewing payoff, audience relationship, distribution response.
Express these ideas in plain language instead:
- “What viewers will think you are about to cover when they see the opening…”;
- “Only … will care about this issue.”;
- “The truly valuable content is later, but many people will never reach it.”;
- “This line will make people stop / swipe away / comment / save.”;
- “This audience is very close to / far from your product.”
输出格式
Output Format
markdown
undefinedmarkdown
undefined一句话结论
一句话结论
{这条内容会吸引谁;大多数人会继续还是离开;最直接的原因}
{这条内容会吸引谁;大多数人会继续还是离开;最直接的原因}
观众看到开头会怎么想
观众看到开头会怎么想
{他们以为后面要讲什么;为什么想继续或离开}
{他们以为后面要讲什么;为什么想继续或离开}
看下去以后能得到什么
看下去以后能得到什么
- {最重要的收获}
- {只有确实存在时再补充第二项}
- {最重要的收获}
- {只有确实存在时再补充第二项}
可能得到什么流量
可能得到什么流量
- 播放:{...}
- 评论/收藏/关注:{只写最突出的行为}
- 商业价值:{这批人与产品或作者定位的距离}
- 播放:{...}
- 评论/收藏/关注:{只写最突出的行为}
- 商业价值:{这批人与产品或作者定位的距离}
最该改的地方
最该改的地方
- {最优先的一项;说明改后会带来什么变化}
- {确有必要时再写}
开头已经让多数人离开时,输出可以缩成三部分:「一句话结论」「为什么开头留不住人」「后面哪项价值应该提前」。不要为了填满模板继续写完整报告。- {最优先的一项;说明改后会带来什么变化}
- {确有必要时再写}
When the opening already causes most viewers to leave, the output may be reduced to three parts: “One-Sentence Conclusion,” “Why the Opening Fails to Hold Viewers,” and “Which Later Value Should Be Brought Forward.” Do not continue filling out the template merely for the sake of completeness.与相邻 Skill 的边界
Boundaries with Adjacent Skills
- :判断观点、理论、因果或经验结论是否成立;本 Skill 不承担这项任务。
dbs-theory-grounding - :用传播学与社会心理学分析共鸣机制、受众情绪和传播动机;本 Skill 侧重受众范围、观看行为与商业价值。
dbs-spread - :用传播学和社会心理学诊断文稿共鸣机制;本 Skill 更关注整体受众范围、观看回报与结果分层。
dbs-resonate - :在选题确定后设计内容表达;本 Skill 先判断内容可能引发哪些受众反应。
dbs-content
- : determines whether a viewpoint, theory, causal claim, or empirical conclusion is valid; this Skill does not perform that task.
dbs-theory-grounding - : uses communication studies and social psychology to analyze resonance mechanisms, audience emotions, and sharing motivations; this Skill focuses on audience scope, viewing behavior, and commercial value.
dbs-spread - : uses communication studies and social psychology to diagnose resonance mechanisms in a draft; this Skill is more concerned with overall audience scope, viewing payoff, and differentiated outcomes.
dbs-resonate - : designs content expression after the topic has been selected; this Skill first determines what audience reactions the content may trigger.
dbs-content
停止条件
Stopping Conditions
- 已回答用户提出的具体问题;
- 已给出受众关系、观看回报、传播反应、商业价值和优先建议;
- 依赖平台、画面、账号或后台数据的部分已标记为未知;
- 用户只要求某个局部判断,已经完成该判断及必要解释。
- The user’s specific question has been answered;
- The audience relationship, viewing payoff, distribution response, commercial value, and priority recommendations have been provided;
- Any parts that depend on platform, visual, account, or backend data have been marked as unknown;
- If the user requested only a focused judgment, that judgment and the necessary explanation have been completed.
自检
Self-Check
- 是否区分了用户指令与文稿内部问句?
- 是否隔离后文,只用开头判断观众以为内容要讲什么?
- 是否检查了开头显露的问题原本存在于多少人心中?
- 开头留不住人时,是否明确说后面的好内容多数人看不到?
- 是否避免用后文资产反向提高停留和后续反应预测?
- 是否检查了观众是否关心冲突对象,而非只识别冲突句式?
- 是否直接分析受众反应,没有转入观点核验?
- 是否明确说出内容与哪些人有什么关系?
- 是否说清观众实际获得的观看回报?
- 是否分别判断停留、持续观看、评论、收藏、转发和关注?
- 是否把片面归因、强断言和身份冲突当作可能的传播元素分析?
- 是否把商业价值落到流量构成、能力展示和后续行动?
- 是否只从受众反应中寻找流量限制?
- 是否说明建议会增强什么、牺牲什么?
- 是否没有预测具体播放量?
- 对外是否没有出现「首屏、闸门、通过、可抵达、观看回报、受众关系、传播反应」等模型术语?
- 是否先说结论,并把默认篇幅控制在用户能快速读完的范围?
完成当前任务后直接结束。只有用户明确询问下一步,且当前环境已经安装 时,简短提示:「下一步不确定时,可以输入 。」
/dbs/dbs- Were user instructions distinguished from questions inside the draft?
- Was the later content isolated so that only the opening was used to determine what viewers think the content will cover?
- Was it checked whether the question raised by the opening already existed in viewers’ minds, and for how many?
- If the opening cannot hold viewers, was it explicitly stated that most people may not reach the valuable later content?
- Were later assets prevented from retroactively improving predictions of stopping and subsequent responses?
- Was it checked whether viewers care about the object of the conflict, rather than merely identifying a conflicting sentence structure?
- Was the audience response analyzed directly without shifting to claim verification?
- Was it made clear which people the topic relates to?
- Was it explained what viewers actually gain from watching?
- Were stopping, continued watching, commenting, saving, sharing, and following assessed separately?
- Were one-sided attribution, strong assertions, and identity conflict analyzed as possible distribution elements?
- Was commercial value tied to traffic composition, demonstrated capabilities, and subsequent actions?
- Were traffic constraints identified only from audience responses?
- Was it explained what each recommendation would strengthen and what it would sacrifice?
- Were specific view counts avoided?
- Were model terms such as “first screen,” “gate,” “pass,” “reachable,” “viewing payoff,” “audience relationship,” and “distribution response” kept out of external reports?
- Was the conclusion presented first, with the default length kept within a range the user can read quickly?
End immediately after completing the current task. Only when the user explicitly asks about the next step and is already installed in the current environment, briefly suggest: “If you are unsure what to do next, you can enter .”
/dbs/dbs