PromptLint

Underlines weak AI prompts as you type in ChatGPT, Claude, Gemini & Perplexity. Live score + 1-click restructure. 100% local. Ever asked ChatGPT to "improve this" and gotten mush back? The problem usually isn't the model — it's the prompt.

PromptLint is a real-time linter for your AI prompts. As you type in ChatGPT, Claude, Gemini, or Perplexity, it underlines the weak spots in your draft before you hit Enter:

• Vague asks with no object or goal
• Missing output format ("in what shape? how long?")
• Dangling references ("fix this" — this *what*?)
• Multi-ask overload (4 questions in one breath)
• Missing context or audience, contradictions ("brief but comprehensive"), token-waste filler, and unbounded open-ended asks

A live Prompt Strength score (0–100) sits at the corner of the composer. Click it for the full issue list with fix suggestions — or hit "Restructure prompt" to reorganize your own words into a clean Role/Context → Task → Details → Output format skeleton, with [ADD: …] markers showing exactly what's missing. Insert it back with one click; Ctrl+Z restores your original. It never sends anything for you.

Private by construction: 100% local rule engine. Zero network calls, zero analytics, no account, and the only permission is "storage" for your own settings. Your prompts never leave the page.

Better prompts in, better answers out.