PromptHub: A Local-First Workbench for Prompts, Skills, and AI Agent Assets
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
At a glance
- What is it?
- legeling/PromptHub is an AGPL-licensed desktop and web application that manages prompts, SKILL.md files, and AI coding assets in a local workspace, then distributes them to a dozen AI development tools with one click. It targets developers who use multiple AI coding assistants and want to maintain a single authoritative source for their prompts rather than duplicating them across tools.
- Who is it for?
- PromptHub makes sense for individual developers or small teams who maintain a growing library of prompts and SKILL.md files and want to push them consistently to multiple AI coding tools without maintaining copies in each tool's config directory. The AGPL-3.0 license means any modified version distributed as a service must share its source, which is a constraint for companies building internal tooling on top of PromptHub.
- Can I use it commercially?
- Yes, with strict conditions. AGPL-3.0 is a network copyleft licence: if people use a modified version over a network, for example as a hosted service, you must offer them its source code under the same licence.
- Is it still maintained?
- Yes. The repository last received commits 3 days ago.
- What is it written in?
- Mainly TypeScript, according to GitHub's language statistics.
Answers come from the project's GitHub data, last synced on October 1, 2026, and from our analysis. They are not legal advice.
Editorial analysis
The Problem: Prompt Files Scattered Across a Dozen AI Tools
A developer who uses Claude Code, Cursor, and Cline alongside each other faces a specific maintenance problem: each tool has its own configuration directory for skills, prompts, and custom instructions. When a prompt is updated, the change must be manually propagated to each tool's directory. When a team standardizes on a new prompt, every developer must update every tool separately.
PromptHub addresses this by acting as a single local workspace that holds the canonical versions of prompts, SKILL.md files, and project-level AI programming assets. A change made in PromptHub propagates to connected tools in one operation. The README describes the core capability as putting your Prompt, SKILL.md, and project-level AI programming assets into one local workspace and distributing the same skill to Claude Code, Cursor, Codex, Windsurf, Antigravity, Cline, and more than ten other tools in a single click.
Data is stored on the user's own machine by default. There is no mandatory cloud backend.
Three Management Domains: Prompts, Skills, and Agents
The application organizes assets into three areas.
Prompt management covers creating, editing, versioning, and testing prompt templates. PromptHub supports multi-model testing, meaning the same prompt can be sent to multiple configured LLM providers and the outputs compared side by side. This is useful during prompt development to verify that a prompt behaves consistently across models a team deploys.
Skill management handles SKILL.md files and similar structured instruction sets used by AI coding tools. The README describes the skill distribution mechanism as the central capability: one installed skill in PromptHub, one click, and the same skill file appears in the configuration directory of every connected tool.
Agent asset management covers what the README calls project-level AI programming assets. The monorepo structure (apps for desktop, web, CLI, and mobile) suggests that this includes CLAUDE.md-style files, agent configurations, and other context documents that AI coding assistants consume from a project's root directory.
All three domains share the same sync and backup infrastructure: WebDAV for cross-device synchronization and a self-deployed web interface for full snapshots.
Downloading and Installing the Desktop App
The primary distribution method is a pre-built desktop application. The latest stable release at the time of the last push (2026-09-24) is v0.5.9, with a beta v0.6.0-beta.2 also available. The GitHub Releases page hosts installers for Windows (x64 and ARM64) and macOS (Apple Silicon and Intel).
The README notes a direct download link that points to the GitHub Latest stable release assets, and a separate GitHub Releases page for accessing older versions, release notes, and signature files.
For developers who want to build PromptHub from source, the repository uses pnpm as its package manager in a monorepo layout managed by Turborepo. The workspace is declared in `pnpm-workspace.yaml` and covers `apps/`, `packages/`, `modules/`, `locales/`, and `tools/`. The desktop application build target is `@prompthub/desktop`.
A CLI is also part of the monorepo. The package.json lists a `build:cli` script, and the repository root contains a `.codex/` directory, indicating some integration with OpenAI Codex-style tooling. The full CLI documentation lives in the truncated section of the README's command-line CLI chapter.
Version Control and WebDAV Sync
Prompt version management means that every edit to a prompt creates a record. This is useful when an update makes a prompt perform worse: the developer can inspect the diff between versions and roll back without relying on git or external versioning.
Multi-model testing integrates version management with provider access. A prompt can be sent to different configured LLM endpoints from within the version comparison view, letting the developer see whether a prompt change degrades performance on one model while improving it on another.
WebDAV sync adds cross-device access to the local-first model. If the user runs a WebDAV server (or points PromptHub at a NAS or a cloud storage service that exposes WebDAV), the local workspace synchronizes across machines. This is a deliberate architectural choice: the data never goes to a PromptHub server, but a user-controlled WebDAV endpoint can make it accessible from multiple devices.
The self-deployed web backup option creates a snapshot of the entire workspace that can be stored on a web server the user operates. This serves as both a backup and a read-only reference for team members who cannot install the desktop app.
What PromptHub Does Not Handle
PromptHub is a management and distribution tool, not an AI inference service. It does not call LLM APIs on behalf of the user during normal operation; the multi-model testing feature requires the user to configure provider API keys, and those calls go directly from the client to the provider.
The project is in active beta development. The latest stable version is v0.5.9, and a v0.6.0-beta.2 has been released but not declared stable. The CHANGELOG exists and releases are relatively frequent, but the version number below 1.0 signals that the API and feature set may change.
The AGPL-3.0 license is a practical constraint for enterprise use. Any software that incorporates PromptHub code and is deployed as a service, including an internal company deployment, must release the source under AGPL-3.0 if modifications are distributed. Teams evaluating PromptHub for internal tooling should review this requirement with their legal team.
The mobile app (`apps/mobile`) exists in the monorepo but is not among the documented installation options in the stable release. It is a work in progress.
Comparison with Manual Prompt File Management
The alternative to PromptHub is managing prompt files and SKILL.md files manually in each tool's configuration directory: keeping a Claude Code SKILL.md in the project root, a Cursor rules file in `.cursor/`, and a Cline instructions file in its own location. Git handles version control of those files.
The trade-off is clear: manual management is simpler, has no licensing constraints, and requires no additional software. The friction is that propagating a change across multiple tools requires editing each config file separately, and there is no built-in multi-model test harness for comparing prompt performance across providers.
PromptHub adds value when the number of tools and prompts grows large enough that the per-tool duplication becomes costly. A developer using two AI coding tools with five prompts each will not see much benefit. A team standardizing 30 prompts across seven tools, with active experimentation on model choice, is closer to the intended use case.
For LangChain users, LangChain Hub is a cloud-based alternative for storing and discovering prompts. The key difference from legeling/PromptHub is that LangChain Hub is a cloud-hosted service, while PromptHub stores everything locally and requires no cloud account.
Release Cadence, Monorepo Structure, and License
The last push was on 2026-09-24. Recent releases show a consistent pattern: v0.5.9 on 2026-07-14, v0.6.0-beta.1 on 2026-08-19, and v0.6.0-beta.2 on 2026-09-03. The project is active and moving toward a 0.6.0 stable release.
The monorepo packages version is `0.6.0-beta.2` in package.json, consistent with the latest GitHub release. The workspace structure covers `apps/api`, `apps/web`, `apps/desktop`, `apps/mobile`, and various packages and modules. The TypeScript toolchain uses pnpm 9.15.0, Turborepo for task orchestration, and Vite for bundling.
The license is AGPL-3.0. This is a copyleft license with a network use provision: if you run a modified version of PromptHub as a service, you must release your modifications under AGPL-3.0. Using the unmodified application for personal or team use does not trigger this requirement. Distributing a fork or a derivative desktop app does.
Editorial conclusion
PromptHub makes sense for individual developers or small teams who maintain a growing library of prompts and SKILL.md files and want to push them consistently to multiple AI coding tools without maintaining copies in each tool's config directory. The AGPL-3.0 license means any modified version distributed as a service must share its source, which is a constraint for companies building internal tooling on top of PromptHub. Teams that work primarily in one AI coding environment and do not need cross-tool distribution will find little advantage over managing their SKILL.md files directly in that tool's configuration folder.
Frequently asked questions
What is PromptHub?
PromptHub is a local-first desktop and web application for managing prompts, SKILL.md files, and AI agent assets. It stores data on your own machine and distributes skills to AI coding tools like Claude Code, Cursor, Codex, Windsurf, and Cline in a single click.
How does PromptHub compare to alternatives?
PromptHub's defining difference from cloud-based prompt libraries is its local-first model: data stays on your machine unless you configure WebDAV sync. It adds value when you use multiple AI coding tools and want to maintain one authoritative copy of each prompt or skill instead of duplicating files across each tool's configuration directory.
What are the alternatives to PromptHub?
The simplest alternative is managing SKILL.md and prompt files manually in each AI coding tool's configuration directory, with git for version control. For cloud-hosted prompt sharing, LangChain Hub is a well-known option, though it is a cloud service rather than a local-first desktop application.
Official sources
Add this badge to your README
If you maintain this project, the badge below links readers to this analysis and shows its maintenance status from the daily GitHub snapshot. Paste the markdown into your README; add ?metric=license or ?metric=stars to the image URL for a different field.
[](https://hysenlabs.com/projects/legeling-prompthub)