Frequently asked questions
Is my conversation uploaded?
No. The transcript is parsed and converted entirely in your browser. Nothing you paste is sent anywhere or stored.
Which formats does it understand and produce?
It reads OpenAI or Anthropic messages JSON, ChatML, the Llama 3 header format, and plain Role: text, detecting the input automatically. It can output any of those formats, chosen with the tabs.
How does it know my input format?
It looks at the content: JSON arrays are read as messages, the ChatML and Llama markers are recognised by their special tokens, and anything else is read as plain lines that start with a role such as User: or Assistant:.
How are system messages handled for Anthropic?
Anthropic keeps the system prompt separate from the messages, so any system turns are merged into a top-level system field and the remaining user and assistant turns go into the messages array.
Will it keep multi-line messages intact?
Yes. In plain text mode a message continues across lines until the next role label, and in the other formats the full content of each turn is preserved when converting.