Offline Prompt Tester

Test any prompt with multiple LLM-powered perturbations using local backends

Offline Prompt Tester lets you stress-test your LLM prompts by generating variations — rephrased, simplified, formalized, or completely restructured — entirely on your own machine. No cloud. No API keys. No data leaves your device. Connect it to any locally-running LLM (Ollama, Jan.ai, LM Studio, or any OpenAI-compatible endpoint) and see how your prompts hold up when reworded. Free — no account needed • 4 perturbation types: Rephrase, Simplify, Formalize, Informalize • Compare up to 3 types side by side • Text similarity scoring (Levenshtein) • Word-level diff view (highlights exactly what changed) • Token count estimate per result • Pin your best results across sessions • Export results as .txt, .csv, or .pdf • Prompt library (save up to 5 prompts) • 7-day history • Right-click any text on any page → "Test this prompt" • Omnibox shortcut: type pt <your prompt> in the address bar • Auto-detects available models from your local server • 10 free tests per day Pro — one-time purchase, lifetime license • Unlimited tests per day • 9 additional perturbation types: Add/Remove Ambiguity, Change Perspective, Reverse Polarity, Add/Remove Constraints, Domain Shift, Role Injection, Adversarial AI similarity scoring and sentiment delta (via your local LLM) • Batch mode: run a perturbation across multiple saved prompts at once, with CSV export • Save up to 50 prompts • 30-day history