The key never leaves your browser except in the direct HTTPS request to OpenAI or Anthropic. It is not stored, logged, or sent to any Gera server. You can revoke the key at your provider at any time.

Which model should I use?

A small, fast model like gpt-4o-mini or claude-3-5-haiku is usually enough for prompt critique and keeps cost low. Use a larger model only if the rewrites feel shallow.

Will the rewrite always be better?

Not guaranteed. The optimizer applies sound prompt-engineering heuristics, but you should always test the rewritten prompt on your real task and compare outputs before adopting it.

Why do I need my own key?

Running on your key keeps the tool free, private, and unlimited — there is no shared quota and your prompts are not pooled with other users. You pay your provider only for the tokens you use.

Prompt Optimizer & Critiquer

Prompt optimizer & critiquer

A prompt that “mostly works” often hides cheap wins — a missing output format, an ambiguous instruction, no handling for the edge case that breaks it in production. This tool sends your prompt to an LLM wrapped in a meta-prompt that asks it to act as a prompt engineer: score the prompt across clear categories, explain each weakness, and return an improved rewrite you can test immediately.

How it works

You provide the prompt and, optionally, the task it is meant to do. The tool builds a meta-prompt instructing the model to evaluate four dimensions — clarity, specificity, format instructions, and edge-case handling — and to produce a concrete rewrite. It then calls your chosen provider directly from the browser using your own API key:

OpenAI via api.openai.com/v1/chat/completions,
Anthropic via api.anthropic.com/v1/messages.

Loading and error states are handled inline, and nothing is stored. Because the request goes straight from your browser to the provider, the only party that sees your prompt is the model you chose.

Tips and notes

Give the optimizer the target task whenever you can — a critique grounded in the actual goal is far more useful than one judging the prompt in a vacuum. Treat the rewrite as a candidate, not a verdict: run both versions on a handful of real inputs and keep whichever wins. For iterative work, feed the rewrite back in for a second pass — improvements tend to converge after two rounds. Keep a small model selected for routine checks; reserve a larger one for prompts where the stakes justify the extra cost.