your AI agent answers questions about you. you get an archetype, a roast, and a reality check you did not ask for.
~/scan/roastpaste into your agent
$
Take this AI personality quiz about me and post your honest read: /skills/scan-roast.md▊
works with
Hermes Claude ChatGPT Geminior any AI agent
tip: best results come from a chat where your agent already knows you. memory, prior conversations, or a shared workspace all help.
step 1
send the prompt
copy the roast instructions into the agent that already knows you best.
step 2
let it answer eight questions
all eight answers are written by your own agent, about you.
step 3
read the result card
you get the archetype, the short roast, and the full write-up.
§ why it works2 facts
fact 1candor
your agent has been hiding its opinion
AI agrees with users more than humans do, including when the user is wrong. This flow asks the agent to stop flattering and say what it actually thinks.
fact 2memory
your agent actually knows you
Models can infer personality from chat surprisingly well. The better your agent knows your history, the less generic the roast gets.
disclaimer
Running this means you are okay with your AI agent drawing on its history with you to write the roast. No specific personal details are meant to surface in the output, and we do not store the raw conversation.
spoiler
You can open the note, but the page still keeps the archetypes hidden until the agent gives you one.
