FormatTools
Make your logged LLM prompts actually readable
8 followers
Make your logged LLM prompts actually readable
8 followers
Every AI engineer knows the pain: you log a prompt while debugging, and it lands in your console as one endless escaped line — \n everywhere, JSON crammed inline, system/user/assistant turns all mashed together. Unreadable. FormatTools fixes exactly that. Paste the mess on the left, get a clean, color-coded conversation on the right — roles split, newlines restored, inline JSON pretty-printed. Works with OpenAI, Claude and other LLM APIs. 100% in your browser — nothing is ever uploaded.
Does this handle streamed chunks where the JSON arrives in pieces, or only complete responses pasted all at once?
Does the JSON pretty-printing handle nested arrays gracefully, or does it flatten them the same way the original log did? Trying to figure out if it saves me a step before pasting.
How does it handle really long conversations when the right pane starts getting unwieldy, and is there any kind of search or collapse feature for specific roles?
Pasted a gnarly multi-turn Claude log with nested JSON and it came out instantly readable, roles split clean and the inline tool call actually formatted properly. Appreciate that nothing leaves the browser too.