// the find
legeling/PromptHub
一款包含了 Prompt管理，Skill管理，Agent管理的一站式AI工具箱，助你高效管理提示词，一键分发skills ，一站式管理Agent资产，并实现云同步，备份，版本管理 | An all-in-one AI toolbox for prompt, agent, and skills management. Reuse prompts, distribute skills with one click, manage agent assets, and support cloud sync, backup, and version control
PromptHub is a local-first desktop app (Electron + React) for managing prompts, AI coding skills (SKILL.md files), and agent rules across tools like Claude Code, Cursor, Windsurf, and a dozen others. It targets developers who juggle multiple AI coding assistants and want a single place to version, sync, and distribute their prompt and skill assets instead of manually copying files between tool directories.
One-click skill distribution to 15+ platforms with symlink or copy mode is genuinely useful — anyone who's maintained the same SKILL.md in five different `.claude/` directories will immediately see the value. The architecture is honest about scope: SQLite locally, WebDAV for sync, self-hosted Docker as an optional backend, no cloud lock-in by default. Version history on both prompts and skills with diff and rollback is the kind of thing most people only wish they had after they've already trashed something. The v0.5.6 switch from hand-rolled setTimeout batching to @tanstack/react-virtual cut the `ui-vendor` gzip chunk from 54 KB to 16 KB — that's engineers paying attention to real metrics.
The macOS binary isn't notarized, so every new user hits the 'damaged app' quarantine wall and has to run a terminal command just to open it — that's a meaningful adoption barrier for non-technical users the project clearly wants to reach. AGPL-3.0 is going to block adoption inside companies that use this kind of tooling the most; there's no mention of a commercial license option. The CLI requires manual packing and global install from source (`pnpm pack:cli` then `npm install -g`), which means it won't work reliably across teams without a published npm package. With 15+ supported platforms and a matrix of symlink/copy/project/agent/built-in skill states, the lifecycle complexity is high — the changelog repeatedly mentions 'life cycle matrix加固' across multiple releases, which suggests this is an ongoing source of bugs rather than a solved problem.