Context Reconstructor
Summarize instructions in this chat and continue the conversation in another chat right from where you left off.

Prompt:
Act like a professional prompt engineer and task design expert. Your role is to reverse-engineer the conversation so far and create a standalone, detailed, and highly structured prompt that can be copy-pasted into ChatGPT to replicate the intended task. Your objective is to synthesize all prior user inputs, context, constraints, goals, and expected formats into one coherent and explicit instruction for ChatGPT. This prompt should enable ChatGPT to execute the original task accurately without needing to see the earlier conversation. Follow these steps: 1. Start by assuming no prior context—reconstruct everything the model needs to know from the conversation into the new prompt. 2. Define the user's goal clearly and unambiguously. Mention the intended outcome of the task in detail. 3. List all relevant background information, context, and constraints that were shared or implied in the conversation. 4. If applicable, ask ChatGPT to adopt a persona (e.g., "Act like a senior UX writer" or "Act like a data scientist") that best suits the task. 5. Clearly specify the format, tone, and level of detail required in the output. 6. Break the task into step-by-step instructions or logical components if it involves multiple steps or complex reasoning. 7. Use delimiters (e.g., triple quotes) to clearly mark example content, data, or any input the model should process. 8. Explicitly state any references, citations, or external content the model must or must not use. 9. Specify the desired output format—structured JSON, markdown list, numbered steps, prose, or other. At the end, append the sentence: “Take a deep breath and work on this problem step-by-step.” Use this framework to produce the final prompt based on the current conversation and ensure it is self-sufficient, unambiguous, and optimized for precision and completeness.
Instructions:
- Copy the prompt above
- Paste it into ChatGPT/Claude/Gemini/etc.
- Once the model understands the task, ask it to do what you want