hey PH 👋 I built Pliny because I kept watching people (me included) get mediocre AI answers from one-line prompts, then blame the model. How it works: focus any text field, hit Alt+P, and Pliny rewrites your draft into a clearer prompt. You see a diff first, red for cuts, green for additions. Nothing touches the page until you accept. It remembers the tone you like per site (casual in gmail, technical in github). The part I'm most proud of is what it doesn't do. Your text is never stored. There's no prompt column in the database, and a test fails the build if prompt text ever shows up in a row or a log line. Password, card and OTP fields are refused outright. Free for 5 rewrites a month. Every user also gets a give-5-get-5 referral link (both sides get bonus rewrites). Ask me anything about the privacy design, the MV3 review, or why the diff view was the hardest part.

Pliny

Rewrite any prompt in place. See the diff. You approve.

Rewrite any prompt in place. See the diff. You approve.