@terreno/ai

AI service layer for Terreno apps - GPT chat, request logging, and admin tools.

AI service layer for Terreno backends — provider-agnostic chat, request logging, GPT history, projects, file uploads, MCP tools, and optional Langfuse integration.

Install

bun add @terreno/ai @terreno/api mongoose

Also install a Vercel AI SDK provider in your app (e.g. bun add @ai-sdk/google).

Quick start

import {TerrenoApp} from "@terreno/api";
import {AiApp, AIService} from "@terreno/ai";
import {google} from "@ai-sdk/google";
import {User} from "./models/user";

const aiService = new AIService({model: google("gemini-2.5-flash")});

new TerrenoApp({userModel: User})
  .register(new AiApp({aiService}))
  .start();

This mounts GPT chat (POST /gpt/prompt), history CRUD (/gpt/histories), projects, and the admin AI request explorer. Add LangfuseApp, FileStorageService, and MCPService when you need those features.

What's included

• AIService — text, streaming, structured JSON, remix, summary, and translation helpers
• AiApp — TerrenoPlugin that registers all AI routes in one call
• Mongoose models: AIRequest, GptHistory, Project, FileAttachment
• Route registrars: addGptRoutes, addGptHistoryRoutes, addProjectRoutes, addFileRoutes, addMcpRoutes, addAiRequestsExplorerRoutes
• LangfuseApp — optional Langfuse tracing and admin UI routes
• FileStorageService — GCS uploads with signed URLs
• MCPService — SSE MCP client for tool calling
• getMCPTools — wrap registered modelRouter MCP tools as Vercel AI SDK tools for in-process streamText / generateText
• Gemini and Vertex helpers: listGeminiApiModels, createVertexProvider

Documentation

Full API reference: docs/reference/ai.md

License and Contributing

Licensed under the MIT License. See CONTRIBUTING.md for contribution guidelines.