Skip to main content
Glama

use_prompt

Execute registered prompt templates to run wizards, workflows, and guided interactions in the Letta system.

Instructions

Execute a registered prompt template. Use this to run wizards, workflows, and guided interactions.

Input Schema

TableJSON Schema
NameRequiredDescriptionDefault
prompt_nameYesName of the prompt to execute (e.g., letta_agent_wizard, letta_memory_optimizer)
argumentsNoArguments to pass to the prompt (depends on the specific prompt)

Implementation Reference

  • The main handler function for the 'use_prompt' tool. It retrieves a registered prompt by name from the promptRegistry, executes it with provided arguments, and returns the resulting messages in both text and structured content formats.
    export async function handleUsePrompt(server, args) { try { const { prompt_name, arguments: promptArgs } = args; if (!prompt_name) { throw new Error('Missing required parameter: prompt_name'); } // Get the prompt from registry const prompt = promptRegistry.get(prompt_name); if (!prompt) { // List available prompts for helpful error const availablePrompts = Array.from(promptRegistry.keys()); throw new Error( `Prompt not found: ${prompt_name}. Available prompts: ${availablePrompts.join(', ')}`, ); } // Execute the prompt handler const messages = await prompt.handler(promptArgs || {}); return { content: [ { type: 'text', text: JSON.stringify( { prompt_name, description: prompt.description, messages: messages.length, preview: messages[0]?.content?.text?.substring(0, 200) + '...', }, null, 2, ), }, ], structuredContent: { prompt_name, description: prompt.description, messages, }, }; } catch (error) { return server.createErrorResponse(error, 'Failed to execute prompt'); } }
  • The tool definition object for 'use_prompt', including detailed inputSchema (requiring prompt_name and optional arguments) and outputSchema (including prompt details and messages).
    export const usePromptToolDefinition = { name: 'use_prompt', description: 'Execute a registered prompt template. Use this to run wizards, workflows, and guided interactions.', inputSchema: { type: 'object', properties: { prompt_name: { type: 'string', description: 'Name of the prompt to execute (e.g., letta_agent_wizard, letta_memory_optimizer)', }, arguments: { type: 'object', description: 'Arguments to pass to the prompt (depends on the specific prompt)', additionalProperties: true, }, }, required: ['prompt_name'], }, outputSchema: { type: 'object', properties: { prompt_name: { type: 'string', description: 'Name of the executed prompt', }, description: { type: 'string', description: 'Description of the prompt', }, messages: { type: 'array', items: { type: 'object', properties: { role: { type: 'string' }, content: { type: 'object' }, }, }, description: 'Messages returned by the prompt', }, }, required: ['prompt_name', 'messages'], }, };
  • Registration of the 'use_prompt' tool handler in the central switch statement within the CallToolRequestSchema handler, dispatching calls to handleUsePrompt.
    case 'use_prompt': return handleUsePrompt(server, request.params.arguments);
  • Inclusion of usePromptToolDefinition in the allTools array for registration with the MCP server.
    usePromptToolDefinition,
  • Import of the handler function and tool definition from the implementation file.
    import { handleUsePrompt, usePromptToolDefinition } from './prompts/use-prompt.js';

Latest Blog Posts

MCP directory API

We provide all the information about MCP servers via our MCP API.

curl -X GET 'https://glama.ai/api/mcp/v1/servers/oculairmedia/Letta-MCP-server'

If you have feedback or need assistance with the MCP directory API, please join our Discord server