Ye Tong’s Original Xiaohongshu Cover | Public Package Integration
This layer integrates a local fitness business; it is not a separate cover methodology. The user has explicitly chosen to place the personal version
into YETONG as-is: see the full original text in
Original Cover Workflow, the visual standards in
Original Visual Language, and the compositions in
Original Layout Templates. The three source documents remain unchanged, and the public invocation short name remains
. Only the methodology text is distributed; no photos, appearance, venue, person profile, or private assets of Ye Tong or any coach are bundled.
Pre-Entry Checks
- Use the path explicitly handed off by the user in this interaction, and first run
python3 scripts/check_profile.py <person-dna.md>
- This module serves prospective members in the city where this coach operates. There must be a specific topic, body copy, or a content summary that clearly states the promise of the body copy; if the request is only “make a fitness cover,” return to topic selection. If there is only a title and the specific content is unclear, follow Step 0 of the original workflow to request only the content, and do not invent a story from the title. Numbers, city, course standards, and service frequency may use only confirmed, publishable facts.
- The six groups of DNA are the basis for selecting material on a post-by-post basis; they are not another visual template. Draw facts from the venue characteristics, core customer segments, verbatim customer statements, identity and service delivery, publishable highlights, and the coach’s own reasons for choosing them when relevant to the current post. Do not treat questionnaire examples as facts, and do not convert self-statements into member testimonials. Use the original visual and Q&A flow; do not replace it with the legacy five fitness modes, a neutral palette, or a poster without a person.
- A title explicitly provided by the user is used verbatim as the text on this cover by default; do not add a “Cover” label. Treat it as no cover copy only when the user explicitly says it is solely for the note’s publication title and separately requests cover candidates. Line breaks and top/bottom segmentation may be arranged, but do not delete or alter any words. If no title is provided, generate three numbered candidates as in the original workflow, handling only this cover and not changing the note’s publication title without authorization. If the content contains unsupported performance guarantees, false prices, medical claims, or unauthorized privacy disclosures, state the specific issue and the evidence that must be verified, then stop formal production. Do not propose an alternative title in the cover module or ask whether to use alternative wording. Hand the task to the title/copy module only when the user separately requests a title change; never silently change the title, replace it with a personal story, or use small-print disclaimers to circumvent the issue.
If the profile does not exist, is invalid, or has not been activated, stop personalization and provide the following message verbatim:
The person’s DNA has not yet been activated. At present, only the generic methodology can be invoked, which cannot enable personalized topic selection, mapping of real experiences, or calibration of personal expression. Complete the “Personal Background Interview” and confirm the profile before the complete Ye Tong content-method engine can be unlocked.
Only when the user explicitly skips the interview may you explain the original generic-version workflow without personal facts; do not use another person’s city, appearance, or story as the user’s own cover, and do not activate the profile on their behalf.
Execute the Original Workflow; Do Not Run the Legacy Method in Parallel
After the checks pass, fully read Original Cover Workflow and actually execute Steps 0–6 in order. Keep the original technology/product examples rather than rewriting them as fitness templates; in a local-business context, fill the content slots only with the user’s genuine training topic, equipment, training plan, or service setting. If there is no physical product, you may select the original “single-person talking-head instructional style”; do not invent a product.
- Ask only one current question at a time, wait for the answer, and then continue, preserving the original order of Step 0 → 1 → 2 → 3 → 4 → 5 → 6. Even if the user first says there is no Image 1, and the composition has not been confirmed, first provide the Step 1 composition choices; only after the composition is selected should you stop at Step 2 to request Image 1. Do not request photos before composition selection. Use the original numbering for composition, auxiliary assets, action/pose, background, and no-title choices. Record information already provided and skip those questions, including a step for which the user has explicitly authorized the system recommendation; do not present a lack of selection as an actual selection. If a title has been provided, do not ask for it again, and do not ask extra questions about the font or title effect.
- Track the current step, selected composition, asset ledger, action/pose, background, verbatim title, and output status. When the user replies with a number or uploads an image, return to the same cover stage; do not restart topic selection or the questionnaire. If the context does not reveal what a number means, verify only which step that number corresponds to.
- Image 1 is the person’s reference-material set; if there are multiple images, record them as Image 1a/1b/1c. Image 2 onward consists of auxiliary assets. Map this logical numbering item by item to actual tool attachments/paths, and do not mistake a second person photo for a product. If no readable image has actually been received, do not claim to have viewed one.
- If person reference images are missing, stop at the original Image 1 upload step. Even if the face will not be shown, a reference to that person’s hands, back, side profile, or another partial view is required; do not generate an unfamiliar face from profile text, and do not substitute an information-only poster and call it complete. Do not scan the photo library to find images for the user.
- In a fitness context, product/auxiliary assets may be understood as actual equipment, a training plan, workflow, or venue assets, but retain the original options. If an exact reference is missing, allow only the generic, unbranded items already defined in the original workflow; do not fabricate equipment models, features, Logos, certificates, ratings, chats, or evidence of service.
- Check the source and intended use of each venue or member image. Anonymity does not imply public authorization. For body-comparison images, use only authorized, genuine originals; do not generate or alter body transformations. You may choose a creative background as specified by the original workflow, but must clearly label it as a creative composite, and must not present a generated background as an actual store, an actual past experience captured on camera, or evidence of results.
- After collection is complete, read Original Visual Language in full and use only the selected template from Original Layout Templates. The original fixes 3:4, a live-action portrait-photography feel, oversized top-and-bottom text, white / warm orange-yellow , soft drop shadows, and a small amount of handwritten annotation. These are design specifications of the selected template; do not claim that they are the original venue brand colors of any particular coach. Do not invoke the legacy or .
Prompt and Image-Generation Continuation
Follow the original delivery order: Solution Summary → Complete Image-Generation Prompt → Text Layout Checklist → Generation Constraints. Include text in the prompt by default; only when the user explicitly requests a text-free base image should you leave a safe area and keep the exact text in the layout checklist. Do not call a plan awaiting additional images a finished deliverable, and do not say that reference images have been transferred merely because text was copied across windows.
This is the only additional continuation action in the original workflow and does not alter the previous Q&A: After delivering an executable complete prompt, if the current tool, reference assets, and authorization are all ready and the user has not explicitly requested generation, ask:
Would you like to complete the cover based on this prompt? Reply “Generate Directly” to generate it, or choose “Continue Optimizing the Prompt.”
If the user explicitly requests only a prompt/no generation, do not prompt again; if the current tool has no image-generation capability or cannot receive the necessary reference images, explain the actual gap, retain only executable prompt-optimization options, and do not promise generation with the current tool. If the user has explicitly requested a final image in the current turn and all conditions are met, proceed with direct creation as in the original workflow and do not ask for the same action again.
“Generate Directly” authorizes image generation only for the current round; it does not authorize publishing personal materials or the note. After rechecking the prompt and assets for this post, use the currently available image-generation capability and check person/equipment consistency, fingers, 3:4, Chinese characters, numbers, place names, and whether the title obscures any key areas. If the text is inaccurate, correct it using available editing/text-overlay capabilities. If correction is impossible, state that it is an unfinished draft; do not call it ready to publish. “Continue Optimizing the Prompt” modifies only the prompt for this post and does not generate an image without authorization. If the user needs to select an adjustment direction, follow the original workflow with one numbered question.
Do not promise views, clicks, inquiries, or conversions. Do not install, update, publish, or modify the Skill as part of the cover task; these actions require a separate explicit request.