Codexe Datapack Generator vs Sandstone

Codexe Datapack Generator creates Minecraft datapacks from plain-English descriptions. It can generate mcfunction files, advancements, loot tables, recipes, item modifiers, predicates, and tags, and sets pack metadata for the selected Minecraft version. Users can ask the AI to edit an existing datapack in place through chat. The generator targets Minecraft 1.20.x and 1.21.x with matching pack formats; older versions may work, though the AI defaults to current syntax. Generated datapacks are intended for vanilla use: they need no mod loader, plugin, or Java runtime, and can be placed in a world's datapacks folder and loaded with /reload. Codexe says the generated source folder belongs to the user and may be shared, sold on PMC market, or used on a realm without royalties. The free plan is billed free forever and refills to 50K tokens 24 hours after the last top-up. Pro costs $25 a month for 3M tokens a month. One-time token packs are also available, including $11, paid once for 1M tokens. It runs on web.

Sandstone is a TypeScript library for building, debugging, and sharing Minecraft datapacks and resource packs. It gives developers command and resource autocomplete, with Wiki documentation for datapack elements such as predicates, loot tables, and advancements. You can group multiple functions, advancements, or loot tables in one file, or keep Minecraft’s one-per-file arrangement. Built-in control-flow tools include if/else, boolean logic, loops, switch statements, and sleep; vanilla commands are also available with build-time and run-time checks. The CLI scaffolds and builds packs, watches for file changes, and can rebuild automatically. Documentation snippets can be edited with a live datapack preview without installing Sandstone. The CLI installs Smithed and Sandstone libraries, and npm-compatible package managers are also supported. Functions can be published to NPM for use in other packs. Sandstone is open source and free, and is listed for Linux, macOS, web, and Windows. Local installation requires Bun and Node.js; the documentation recommends a code editor such as Visual Studio Code.