Magic Potion

Visual drag-and-drop prompt editor for crafting, organizing, and reusing prompts across OpenAI, Anthropic, and Google models.

FreemiumWritingMulti-model

Visit website →

Best for

Pick Magic Potion if you write a lot of prompts and want a visual editor and shared library instead of scattered text files and chat tabs.

Skip if

Skip it if you need agent orchestration, RAG, evals, or a programmatic API - this is a prompt authoring tool, not an LLM platform.

Magic Potion is a browser-based prompt engineering workbench that turns prompt design into a visual, drag-and-drop exercise instead of a wall of text. You assemble blocks of instructions, context, and variables into reusable 'prompt stacks,' preview their output against the major LLM providers (OpenAI, Anthropic, Google), and save them to a library you can iterate on or share.

It's pitched at creators, marketers, and prompt-heavy power users who are tired of pasting half-remembered prompts between ChatGPT tabs and Notion docs. The differentiation is the visual editor and the cross-provider preview, which matters if you A/B prompts between Claude and GPT-4 class models. It's a hosted SaaS with a register/login flow at my.magicpotion.app; pricing is not surfaced on the public marketing page, suggesting a freemium-style account model where you sign up to see plans.

It's not a model itself and not an agent runtime, so don't expect tool-use, RAG, or production deployment hooks out of the box. Treat it as a prompt IDE and library rather than an orchestration layer.

Editor's take

A tidy little prompt IDE that scratches a real itch for people who treat prompts as artifacts worth versioning. The visual editor plus cross-provider preview is the right product idea; the open question is whether the library and collaboration features can earn a paid subscription against free alternatives like the OpenAI Playground or Anthropic Workbench.

— The AI Tool Bible editorial team

Pros

✅ Visual block-based editor is faster than maintaining prompts in plain text files
✅ Multi-provider preview across OpenAI, Anthropic, and Google in one UI
✅ Library/stack model encourages reusable, versioned prompts
✅ Clean, focused scope - it does one thing