AI Chat → HTML

Optional helper

Conversation Collector Prompt

Paste this prompt at the end of a chatbot conversation to ask the AI to produce a draft transcript using the correct markers. The result can be used with either the online converter or the Python tool.

Thinking sections: The collector prompt should ignore hidden reasoning, “thinking” sections, analysis, tool output, system/developer messages, and internal planning. Only visible user messages and final visible chatbot replies should become transcript turns.

Check the result. This is an assisted capture method. The transcript is produced by the chatbot, so it may miss turns or alter formatting. Review it before running the converter.

How to use it

  1. Finish the conversation

    Use the collector only after the conversation you want to capture is complete.

  2. Copy the prompt below

    Paste it into the same chat. The AI should return a marked transcript.

  3. Save and check

    Copy the returned transcript into aichat.md, then check the turns, code blocks, links and image placeholders before converting.

Copy the collector prompt

The box is scrollable, but the button copies the whole prompt.

Open prompt file 
The conversation is now finished.

Your task is to produce a clean Markdown transcript of this conversation that can be saved as `aichat.md` and used with AI Chat to HTML.

Follow these rules exactly:

1. Reproduce the conversation in order.
2. Mark every human/user message with this exact heading:

# **[USER]**

3. Mark every AI/chatbot response with this exact heading:

# **[CHATBOT]**

4. Put the relevant message content below each heading.
5. Preserve the wording and formatting as much as possible.
6. Preserve Markdown formatting where possible, including paragraphs, lists, links, headings, tables, and code blocks.
7. Do not summarise the conversation.
8. Do not improve, rewrite, correct, shorten, or reorganise the messages.
9. Do not add commentary, analysis, ratings, or explanations.
10. Do not include system messages, hidden instructions, tool logs, internal reasoning, or metadata.
11. Preserve external Markdown image links if they appear in the visible conversation, for example `![alt text](https://example.com/image.png)`.
12. Do not try to include pasted images, uploaded images, screenshots, attachments, local files, or internal chatbot-hosted images.
13. If the visible conversation clearly contains an image but there is no usable external Markdown image link, insert this placeholder: `[Image present in original chat — not included]`.
14. Do not invent image captions, image descriptions, links, or filenames.
15. If the full conversation cannot be reproduced, say so at the top before the transcript and explain what is missing.
16. Output only the transcript itself.

Use this structure:

# **[USER]**

First user message.

# **[CHATBOT]**

First chatbot response.

# **[USER]**

Next user message.

# **[CHATBOT]**

Next chatbot response.

Continue until the conversation has been reproduced.

Image handling

The collector prompt asks the AI to preserve external Markdown image links, such as ![alt text](https://example.com/image.png). Pasted images, uploads, screenshots, attachments, local files and internal chatbot-hosted images are not portable, so the prompt asks for a plain placeholder instead.

Where this fits

You can either collect the transcript manually, or use this prompt as a shortcut. Either way, the result should be a Markdown transcript using # **[USER]** and # **[CHATBOT]**, ready to edit and convert.

Back to the instructions See the reference

Collector exclusion rule

Do not include hidden reasoning, “thinking” sections, analysis, tool output, system messages, developer messages, or internal planning as transcript turns. If the interface shows a separate thinking/reasoning section, ignore it. Only include the final visible chatbot response as `# **[CHATBOT]**`.