Frequently Asked Questions about Claude Custom Prompts MCP Server

Q: What is the Claude Custom Prompts MCP Server? A: It’s a Node.js server implementing the Model Context Protocol (MCP) for Claude AI models, allowing you to define and use custom prompt templates with a modular, category-based organization system.

Q: What is MCP? A: MCP stands for Model Context Protocol. It’s an open protocol that standardizes how applications provide context to LLMs.

Q: What are the key features of the MCP Server? A: Key features include easy integration with Claude, custom prompt templates, support for prompt arguments with validation, organized prompt categories, multiple transport options, conversation history access, prompt chains, and distributed prompts configuration.

Q: How do I install the MCP Server? A: You need Node.js v16 or higher and npm or yarn. Clone the repository, navigate to the ‘server’ directory, run npm install, npm run build, and then npm start.

Q: How do I use custom prompts in Claude Desktop? A: Once the server is running and configured in Claude Desktop, type >>command_name argument1 argument2.

Q: What are prompt chains? A: Prompt chains allow you to define and execute a sequence of prompts, with each prompt in the chain using the results from previous prompts, enabling complex tasks to be broken down into smaller steps.

Q: What are the special placeholders I can use in prompt templates? A: You can use {{previous_message}} to insert a prompt asking Claude to reference the previous message in its context.

Q: How do I configure the MCP Server in Claude Desktop? A: Locate your claude_desktop_config.json file and add a mcp_servers section with the correct command, arguments, and working directory (cwd). Ensure the paths are absolute and use double backslashes in Windows.

Q: What do I do if the server can’t find configuration files or prompt files? A: The working directory is likely incorrect. The server logs its working directory at startup – check this to verify it’s what you expect. Ensure the cwd parameter in Claude Desktop is correctly configured.

Q: How do I create a new prompt category? A: Create a new folder in the prompts directory, create a prompts.json file in the new category folder, add your category to the categories array in promptsConfig.json, and add the path to your category’s prompts.json file to the imports array in promptsConfig.json.

Q: What built-in commands does the server support? A: The server supports /listprompts (or >>listprompts) to list all available commands, and /command_name [arguments] (or >>command_name [arguments]) to execute a command.

Q: How can I reload prompts without restarting the server? A: You can use the reload_prompts tool via the API or Claude interface. For example, use the command >>reload_prompts in Claude, or make a POST request to /api/v1/tools/reload_prompts.

Q: How do I update or delete prompts via the API? A: Use the /api/v1/tools/update_prompt and /api/v1/tools/delete_prompt API endpoints respectively, sending a POST request with the required parameters in JSON format.

Q: Where can I find the server logs for troubleshooting? A: The server logs are located in the server.log file within the server directory. The exact path depends on your operating system and installation location.

Q: How does the Claude Custom Prompts MCP Server integrate with UBOS? A: UBOS can orchestrate AI Agents, connect them with enterprise data, build custom AI Agents with your LLM model and Multi-Agent Systems. The MCP Server provides custom prompts for those agents.