canmyagent

Submit

Share a prompt you actually use

We read it for injection and exfiltration patterns, swap personal details for blanks anyone can fill in, and give it a link you can text to anyone.

Optional. We will write one if you leave it blank.

One sentence, in plain words.

Write {{like_this}} around anything someone else would need to change.

Blanks found

Optional. Where you found it, so the original author gets credit.

We pick the category and the blanks for you, then run two safety checks. A prompt that passes both gets a Verified mark. One that does not still publishes, unmarked, unless it looks like it is trying to hijack someone’s agent.