GET STARTED / INTRODUCTION

Introduction

promptLM is an open-source framework for managing, versioning, and releasing LLM prompts as code.

Prompts live in a Git-backed repository as structured PromptSpec files. promptLM gives you a CLI, a local web Studio, a REST API, and client SDKs to author, release, and consume those prompts from your applications.

Note

promptLM is in active development. The main branch may be force-pushed, so pin a published release tag when you embed the CLI or SDKs into your own workflow.

The moving parts

  • promptLM CLI (promptlm-cli) — a Spring-Shell command-line tool for creating, releasing, and inspecting prompts. See CLI reference.

  • promptLM Studio — a local web UI launched via the CLI; defaults to http://127.0.0.1:8085/. See Studio.

  • REST API — Spring MVC endpoints under /api/prompts, /api/store, /api/llm, /api/capabilities — used by the Studio and available for integrations.

  • Client SDKs — Java, TypeScript, and Python libraries that load released prompt JSON bundles in your application. Maintained in the separate promptLM/promptlm-clients repository.

  • macOS menubar client — native Swift app that pastes rendered prompts at the cursor via a global hotkey. Maintained in promptLM/promptlm-macos.

Where prompts run

promptLM routes execution through Spring AI by default (OpenAI and Anthropic). An optional LiteLLM gateway can be enabled when you need an OpenAI-compatible proxy in front of additional providers.