Thinking Process: 1.
Analyze the Request: *
Task: Write a product description for an app that integrates with ChatGPT (Wait, the prompt says "These instructions will be used for an app that integrates with ChatGPT" in the system instruction block, but the actual user input is about a dress. I need to follow the user's specific request regarding the *dress* while adhering to the formatting constraints provided in the prompt). *
Correction: Re-reading the system instruction block vs the user input. The system instruction says "Write a 100–150 word product description...". However, the *user* input provides specific product details and a larger context suggesting a standard e-commerce description (150-350 words usually per the expert writer section within the prompt). Wait, there are conflicting constraints. * System Instruction (top): "Write a 100–150 word product description using a 1-sentence hook followed by 4 concise bullet points..." AND "Use second-person voice. Never imply that we manufacture the product. Do not mention where the product was made... Do not include words such as High Concerned or Hign Concerned." * User Prompt (bottom): "Use only the product details below. If a detail is missing, omit it rather than guessing." + "Objective: write a unique, accurate description... roughly 150-350 words unless the instructions request otherwise." + "Format rules... Return only the description HTML." *
Conflict Resolution: Usually, the most recent or specific instruction overrides general system instructions, but the system instruction explicitly sets a constraint (100-150 words, 1 sentence hook, 4 bullets). However, the user prompt also contains a very detailed set of "Writing principles" claiming "Aim for roughly 150-350 words unless the instructions request otherwise." The top system instruction asks for 100-150 words. I should prioritize the specific constraint given at the very beginning of the conversation turn if it contradicts general persona instructions, OR check if the "User Input" contains its own constraints. * Actually, looking closely at the prompt structure: * Top block: "Write a 100–150 word product description using a 1-sentence hook followed by 4 concise bullet points highlighting benefits and key details. Use second-person voice. Never imply that we manufacture the product... These instructions will be used for an app that integrates with ChatGPT." * Bottom block (User Input): Contains a massive "Expert English language e-commerce product description writer" persona block which says "Aim for roughly 150-350 words". *
Decision: The prompt starts with specific constraints ("Write a 100–150 word..."). Then it continues with a long "You are an expert..." block. Often, the initial instruction sets the hard constraint for the output format (hook + 4 bullets), while the rest provides style guidance. However, the user prompt *also* repeats the expert writer instructions. Let's look at the very first line: "Write a 100–150 word product description using a 1-sentence hook followed by 4 concise bullet points..." This seems to be the primary directive for *this specific turn*. The rest looks like meta-instructions copied into the user prompt. I must adhere to the 100-150 word count and the 1-sentence hook + 4 bullets format requested in the very first paragraph of the text provided. *
Wait, re-evaluation: The prompt is structured as a system instruction followed by user input. The system instruction says "Write a 100–150 word product description...". The user input *also* says "Aim for roughly 150-350 words". This is a conflict. *
Safety/Constraint Check: The "System Instruction" header at the very top of the *conversation* (which I am simulating) dictates the task. The user input contains a copy-paste of a prompt template that includes conflicting word counts. Usually, the explicit instruction at the start of the prompt ("Write a 100–150 word...") takes precedence over the generic persona instructions inside the user message body because it defines the immediate task scope. *
BUT, looking at the very bottom of the user message, it says "Deliver the output in English language." and provides specific format rules about HTML tags. *
Crucial Detail: The first paragraph says "Write a 100–150 word product description using a 1-sentence hook followed by 4 concise bullet points...". This is a specific constraint for *this* output. The later text says "Aim for roughly 150-350 words". I will follow the 100-150 word constraint as it is more specific to the structure (hook + bullets). *
Wait, another conflict: The system instruction says "Never mention where the product was made, manufactured, or ships from". The product details say "CN: Guangdong" and "Origin: Mainland China". I must exclude these. *
Another Conflict: The system instruction says "Do not include words such as High Concerned or Hign Concerned". The product details say "High-concerned chemical: None", "Is High Concerned chemical included?". I must exclude this exact phrasing. *
Formatting: First character must be content. No headings at the top. Only allowed HTML tags. Video embeds must be preserved exactly. *
Video: There is one `
