Lore: Native macOS Playground for LLM Prompt Engineering

Lore reimagines what a local LLM playground can be by building it natively for macOS with Cocoa, SwiftUI, and SQLite under the hood. Instead of losing track of prompt iterations in scattered chat windows, Lore automatically versions every generation, letting you time-travel back through your prompt history, compare variants, and mark the outputs worth keeping. It's built specifically for the messy, iterative reality of prompt engineering rather than treating prompts as disposable text. Beyond version control, Lore adds practical engineering tooling: full-text search across your entire prompt library, model-cost awareness so you can see what each experiment is actually costing, and combinatorial runs that let you test multiple prompt variables at once. It supports custom endpoints and local models alongside major hosted LLMs, plus quality-of-life features like tables, syntax highlighting, LaTeX rendering, notes, export options, keyboard shortcuts, Vim mode, and sandboxing for safer experimentation. The result is a tool that feels less like a chat client and more like an IDE for prompts — purpose-built for researchers and developers who iterate constantly and need a reliable record of what worked, what didn't, and what it cost to find out.

1. Download and install Lore on a Mac running macOS 13 or later. 2. Open the app and connect to your preferred LLM provider, custom endpoint, or local model. 3. Write and run prompts, using combinatorial runs to test multiple variables at once. 4. Review generated variants, mark the notable ones, and use time travel to revisit earlier versions. 5. Search your full prompt history with full-text search when you need to find past work. 6. Monitor model costs as you experiment, and export or annotate results with notes for future reference.