Prompt Manager

What is Prompt Manager?

Prompt Manager is a standalone system for managing AI prompts with composable elements. It provides a web interface and RESTful API for creating, organizing, and editing prompts, and integrates with the Model Context Protocol (MCP) for advanced prompt workflows. It targets users who need efficient, keyboard-driven prompt management.

How to use Prompt Manager?

Clone the repository, set up the environment using the provided Taskfile (recommended) or Makefile, then run the startup script. Access the web UI at http://localhost:8081 or use the REST API endpoints programmatically. For real-time editing, connect via WebSocket to /api/prompts/ws/{prompt_id}.

Key features of Prompt Manager

• Create and manage standard and composite prompts
• Compose templates with prompt inclusions and expansions
• Real-time editing via WebSocket
• Access prompts via RESTful API
• Organize prompts in directories
• Integrates with Model Context Protocol (MCP)

Use cases of Prompt Manager

• Managing a library of reusable prompt templates for AI interactions
• Building composite prompts that include and expand other prompts
• Editing prompts collaboratively in real time via WebSocket
• Automating prompt retrieval and updates through the REST API
• Leveraging MCP integration for Claude Desktop compatibility

FAQ from Prompt Manager

How do I install and run Prompt Manager?

Clone the repository, then set up the environment using task setup:py, task setup:js (or make setup), and run ./bin/start_prompt_manager.sh. Open http://localhost:8081 in a browser.

Where are prompts stored?

Prompts are stored in the project data directory (prompt_manager/data/prompts/) and the user config directory (~/.prompt_manager/prompts/).

What transports and APIs are available?

Prompt Manager offers a RESTful API for CRUD operations on prompts and directories, a WebSocket interface for real-time editing, and integration with the Model Context Protocol (MCP) for advanced workflows.

Does Prompt Manager require authentication?

The README does not mention any authentication mechanism for the web interface or API.

What are the runtime dependencies?

The setup requires Python and JavaScript environments, managed via a Taskfile or Makefile. The application runs on localhost port 8081 by default.