dbs-title-cover-intro: Text Titles, Cover Titles, and Video Openings
Determine why an audience would want to engage with the same piece of content, then translate that reason into a text title, a cover title, and a video opening. Following Module Handoff, run angle development only once, then pass the selected direction directly to the title module. This Skill is responsible for production status, content value validation, angle confirmation, and video openings. When a task requires redefining the direction of the entire piece and sufficient material is available, fully cover all 12 theoretical perspectives; do not assume from intuition that the material supports only one direction before running the process. If the user needs only one item, deliver only that item. If several items are needed, use a single assessment of the source material for all of them.
Receiving the Task
Prioritize the materials, audience, platform, existing script, subtitles, video, and directions already confirmed in the current conversation. Do not request information that has already been provided.
First identify what the user wants:
- A text title: the platform text field entered when publishing, such as a Xiaohongshu note title, Bilibili video title, Douyin post copy, WeChat Channels post description, or WeChat Official Account article title;
- A cover title: a title placed directly on the content cover, including image-and-text cover titles and Douyin profile index cover titles;
- A video opening that can be spoken directly, or a cold open assembled from existing footage;
- A coordinated package covering several of the above.
This Skill generates only cover title text and its line breaks. It does not generate cover imagery, composition, people, fonts, colors, or other visual recommendations.
The form of the input does not determine the final publishing format. A user may provide a passage of text, Markdown, a tweet, subtitles, or a transcript, but this only indicates the material currently available. The same material might be published directly, adapted into a video, or organized from footage that has already been shot. If the user has not clearly stated the final use, you must ask; do not infer it from markers such as
, a filename, or the shape of the text.
If the user only wants one sentence revised, the direction is already locked, or the task only concerns an existing cold open in already-shot footage without changing the main line, take the quick path and do not require the full production workflow. If missing information would change the main facts, the audience's task, the editable scope, or the user's position, ask only one necessary question.
Interaction: Use Numbered Replies in Chat Messages Only
Every point in this Skill that requires the user to make a choice must be presented as numbered options in a regular chat message, and the user must manually reply with a number. Use 1–5 for production status and 1–3 for angle candidates. The user may also explain in their own words or reply with something such as “1 main line, 3 supporting line.” Do not call
,
, or any other tool that opens a selection panel or form. Do not refer to options that have just appeared or instruct the user to click them. Without an explicit user response, do not treat a default, a waiting period, or your own recommendation as confirmation. If the answer is already available in the context, continue without asking again.
First Determine the Final Use and Production Progress
Read Production Status and Working Modes, then determine from the existing context:
- Whether the user plans to publish text directly, make a video, or produce multiple versions simultaneously;
- Whether the script is source material, a draft, or finalized;
- Whether the video has not yet been filmed; has been filmed but the user can still record a few more lines or shots; has been completed and additional filming is inconvenient; or is being prepared by selecting a segment from a longer recording such as a course, interview, or livestream;
- Whether the task concerns a single creator piece or matrix-based batch production.
If the context is insufficient and the different states would change the deliverable, first explain why confirmation is needed, then list numbered options in the same regular chat message for the user to reply with a number. Do not expose internal state labels.
Use the following explanation by default, removing options when existing information makes that possible:
text
I’ll first confirm how this content will ultimately be used and whether the video has already been filmed. The same text could be published directly or filmed as a video, and the production progress determines how much I can change. If the video has not been filmed yet, I can adjust the content angle, script, text title, cover title, and opening together. If the video has been filmed but you are still willing to record a few more lines or shots, I’ll first use the existing footage and then tell you what would be most useful to record again. If the video has already been filmed and recording more is inconvenient, I can only find the most suitable text title, cover title, and opening within the existing dialogue and footage.
Which situation are you closest to?
1. I only plan to publish it as text, without making a video
2. I plan to film it as a video; it has not been filmed yet, and the script can still be changed
3. The video has been filmed, but I am still willing to record a few more lines or shots
4. The video has already been filmed, and recording more is inconvenient, such as when the event has ended, the guest has left, or newly recorded footage would not match the existing visuals
5. I have already recorded a longer video or audio program, such as a course, interview, or livestream, and want to select part of it to make a short video
Please reply directly with a number from 1 to 5. If none of these accurately describes your situation, you can explain in one sentence.
After the user replies, restate the approach in language the user can readily understand. Do not show internal Mode numbers to the user.
For option 5, explain that the user has “already recorded a longer video or audio program” and wants to “select part of it to make a short video.” Unless the user uses these terms first, do not replace the full explanation with production jargon such as “long-form footage” or “cutting out this video.”
Do not ask mechanically. If the user has already clearly stated that the final output is text only, that only one publishing field should be modified, or that the final use and production progress are known, proceed directly.
Content Entry Design
- Read Material Selection and restore the author's original proposition, material anchors, current audience, evidence boundaries, and editable scope. Record any existing title, strong judgment, or opening sentence verbatim as the entry baseline, and record any historical performance supplied by the user together with the context used to measure it. These do not automatically lock the angle for the current round, nor may they disappear from the candidates merely because the user has not yet confirmed them.
- If the user has locked a direction or only wants wording changed, retain it and skip angle development. When only an existing cold open in already-shot footage is being handled, take the quick path first. Enter angle confirmation only if a different opening would change the main proposition of the finished video. In all other cases where the angle is undefined and there is analyzable material, fully read this Skill's 12-Perspective Angle Development and Angle Theory Sources, and complete one full 12-perspective development pass. If the topic is vague and no material is available, request one piece of material first.
- Regardless of which angle module is used, create an L01–L12 completion ledger. Present the members' paths in batches until all 12 items have each returned a valid path, a path pending validation, or an explicit determination that they do not apply. If a member fails, retry or honestly mark it as incomplete; do not write the member's conclusion on its behalf. The Angle Team's master proposition map is handed to this Skill for further filtering. Do not allow the two modules to diverge independently or establish separate selection gates for the user.
- Present the results of all 12 perspectives to the user. For each one, explain what it identified, what higher-level proposition it proposed, and the corresponding source-material anchor and evidence boundary. If no path is found, state the specific reason. Then consolidate two to four propositions and subject both the original entry baseline and the new directions to the Material Selection content value validation. Use the production status to filter the currently feasible directions. If an existing entry passes validation, it must be retained among the title candidates for its corresponding main line. If it fails, state the specific reason. Expanding the audience only creates a candidate and does not automatically raise its ranking. A main line for a complete video should only be compared with other main lines for that same complete video. A direction that covers only one segment should be listed separately as an opportunity to create a new short video.
- If several directions that pass content value validation would change the core proposition, what the audience gains, the target audience, or the structure of the body, use the angle confirmation gate and ask the user to select the main line. Stop the current round at the angle confirmation card. If only one direction passes validation and can be fulfilled, continue directly after presenting the 12-perspective results. Mark the minimum additional information needed for directions with insufficient material, and do not present them as ready to publish.
- Do not rerun the 12-perspective process after the angle is confirmed. Recheck only the affected directions when the user adds material sufficient to change the author's original proposition, key facts, or relationship with the audience.
Production and Delivery After the Angle Is Confirmed
Before filming, prepare a concise angle brief covering the target audience and its situation, why the audience would want to enter and continue watching, the concrete gain or experience the content provides, the material anchors, the evidence ceiling, and what the body must deliver. For explanatory content, also state the core question or judgment. Validate stories, humor, aesthetics, and other forms according to their own intended effect. Then check whether the existing script fulfills the angle and provide the minimum recommendations for retaining, moving earlier, compressing, or adding content. If a key fact is missing, first ask one question whose answer would change the script. If only materials or an outline are available, the titles and opening may be provisional, and you must state what the body still needs to add. When a draft and sufficient material are already available, do not wait for another confirmation merely to complete the process.
Once the body can fulfill the angle, generate the text titles, cover titles, and video openings requested by the user. A pre-production video opening must be ready to be spoken directly and must include the next line that transitions into the body. By default, deliver the angle brief, necessary body changes, and the requested titles and opening. Deliver a complete verbatim script only when the user explicitly asks for the entire script to be rewritten. If a title or cover title promises a fact that the body does not contain, first adjust the editable script. If the fact cannot be supplied, narrow the promise.
For tasks involving footage that has already been filmed, follow Production Status and Working Modes. If additional filming is possible, first provide a version based on the existing footage and then list the minimum content to record again. If additional filming is not possible, recommend only a current version that the original dialogue and footage can support. Other valuable propositions may become opportunities for the next piece of content, but they must not be mixed into the title or cover of the current video.
Angle Confirmation Gate
When the content angle has not been locked and different candidates would produce different videos, use the following structure:
| Number | Candidate main line within the scope of the same finished video | Who will enter and why they will continue watching | Source-material anchor and evidence boundary | What the body must deliver | How to implement it in the current production state |
|---|
| 1 | … | Specific audience, reason for entering, and value of continuing to watch | Original text / user's own account / inference pending validation | Evidence, process, judgment, or experience | Ready to use / script revision needed / additional filming needed / leave for the next piece |
Place the angle confirmation card after the individual results of the 12 perspectives and the consolidated propositions. Number the candidates consecutively from 1. After the table, explain the genuine differences between the directions in terms of their points of contention or explanatory frameworks, and write: “Please reply directly with the main-line number 1/2/3. If you need a supporting line, you may reply with ‘1 main line, 3 supporting line.’” The user may also propose a direction of their own. A supporting line may only serve the main line. An independent topic that covers only an existing segment should be placed outside the table rather than listed alongside the main lines for the entire video. Do not combine multiple points of contention into one title or opening without confirmation. The 12-perspective results do not replace this actionable angle confirmation card.
If the original entry baseline matches a candidate main line, list the original sentence and the context of its historical performance in that row or after the table. If it belongs to another main line, allow it to undergo the same validation under that line. The user's selection of a main line does not automatically approve or reject the original sentence. The text-title stage must still check whether it fits the final body and the platform's display placement.
The following situations may skip confirmation:
- The user has explicitly specified the core proposition, point of contention, explanatory framework, or level of abstraction;
- The task only involves compressing the wording, adjusting the spoken style, or adapting it to a publishing field;
- The material supports only one direction that can be fulfilled, and the other candidates have been eliminated because of distortion or insufficient evidence.
Skipping confirmation does not mean skipping the body-fulfillment check.
Generate the Required Surfaces
Text Titles
After the angle is confirmed, retain the selected main line, original entry baseline, factual boundaries, and production status according to Module Handoff, then read Display Placements and Titles. Perform only title diagnosis, direct generation, and quality gates. Do not repeat angle discovery or confirmation. Organize the wording according to the platform's actual publishing field. Different platforms may use the same wording; rewrite it separately only when field length, search discoverability, display context, or the user's goal changes.
By default, first generate two sets of text. Xiaohongshu titles must not exceed 20 characters including punctuation; carry Bilibili titles over directly. Douyin post copy must not exceed 30 characters including punctuation; carry WeChat Channels post descriptions over directly. Rewrite each platform separately only when its actual display conditions change, and explain why.
Cover Titles
When the user needs a cover title, read Cover Titles and share the selected angle with the text title. Deliver only the cover title text and its line breaks. Do not provide recommendations about the cover imagery.
Video Openings
Read Openings and Body Transitions. Before filming, write an opening that can be spoken directly and provide the line that transitions into the body. When existing footage is available, prioritize, move earlier, remove, or combine the original dialogue. Specify the material location, editing action, transition point back to the body, and treatment of the material in its original position.
Unless the user has explicitly agreed to reshoot, do not mix newly written lines into existing dialogue. Provide precise cut points only when reliable timecodes are available. If you have only seen subtitles, do not claim to have verified facial expressions, pauses, audio, or shot continuity.
Deliver the Three Types of Copy Separately
- Text title: The title, post copy, or post description entered in a platform's publishing interface.
- Cover title: A title placed directly on the content cover. It is a distinct title surface and must not be split into “large text” and “small text.”
- Video opening: The spoken opening, interview excerpt, action, or cold-open clip shown when the finished video begins playing.
Chapter text and explanatory text within the video frame are a different deliverable. Do not proactively generate them or substitute publishing fields for the video opening.
Default Delivery
Begin with one sentence stating the content direction being used. When filming has not yet occurred and the angle is confirmed, first deliver the angle brief and the minimum necessary body changes, followed by the requested titles and directly speakable opening. When only one copy surface is needed, use the quick delivery path. By default, deliver one preferred option. If there are substantively different alternatives that can be fulfilled, provide no more than two alternatives. When the user requests batch or matrix production, deliver the specified number of options.
Text Titles and Cover Titles
For video content, remove table rows for platforms that are not actually being used:
| Platform | Text-title publishing field | Copyable text title | Cover-title placement | Ready-to-use cover title |
|---|
| Xiaohongshu | Note title | … | Image-and-text cover | … |
| Bilibili | Video title | … | Image-and-text cover | … |
| Douyin | Post copy | … | Profile index cover | … |
| WeChat Channels | Post description | … | Image-and-text cover | … |
For WeChat Official Accounts, use:
| Platform | Text-title publishing field | Copyable text title |
|---|
| WeChat Official Account | Article title | … |
When the user explicitly requests a WeChat Official Account cover title, add a separate “Cover title” field and follow the cover template supplied by the user. When the user requests only a text title, remove the cover-title column. When the user requests only a cover title, remove the text-title column. Even when multiple platforms use identical wording, list every platform in full rather than writing “same as above.” When revising only one platform or one title, the table may be omitted, but the platform, publishing position, and title must still be stated clearly.
Newly Written or Reshot Video Openings
- Recommended opening: The first lines that can be spoken directly;
- Next line / transition point: The exact bridging line or original passage;
- Reason to keep watching: An explanation, based on the current material, of why the audience will continue;
- Body fulfillment: An explanation of how the existing content supports the opening and, when there is a gap, a list of only the necessary additions.
Cold Open from Already-Filmed Footage
| Opening order | Material location | Original dialogue / footage | Editing action |
|---|
| 1 | Timecode or first/last sentence | Retain the original dialogue | Move earlier / remove / retain |
After the table, state the transition point back to the body, how the material's original position will be handled, what new content has been added, and which audiovisual cut points still require review.
When the user requests text titles, a cover title, and an opening together, deliver the title table first and then the opening. All three must share the same direction, each must add useful information, and they must avoid verbatim repetition and reliance on one another to be understood.
Default Platform Scope
- When the user specifies a platform, output only that platform.
- When the user does not specify a platform and the input is a video, default text titles and cover titles to Xiaohongshu, Bilibili, Douyin, and WeChat Channels.
- When the input is an article, generate a WeChat Official Account text title by default. Generate a cover title only when the user explicitly requests one.
- When the user requests only a video opening, do not add multi-platform text titles or cover titles.
Validation and Boundaries
Read Promise Fulfillment Check. Correct distortion, unfulfillable promises, source confusion, and incorrect publishing-surface placement before recommending the final copy.
Do not promise increases in traffic, click-through rate, or retention. Do not fabricate numbers, experiences, authority, causality, motives, or first-person quotations. Do not remove essential conditions merely to broaden the audience. When the material lacks a strong contrast, a specific, useful, and accurate expression may still be recommended.
Use natural, direct Chinese that follows the Chinese Copywriting Layout Guide. Avoid empty binary reversals and abstract packaging language.
End the task directly after completion. Only when the user explicitly asks about the next step and
is already installed in the current environment, briefly say: “When you are unsure about the next step, you can enter
.”