Skip to main content
Version: 3.0

Chat App

The Operaide Chat App (core-operaide-chat) is a Core App that ships with every Operaide instance. It provides a conversational AI interface with web search, file uploads, prompt templates, and branding options.

This page walks through installation and configuration. The same general pattern applies to any App, but the Chat App makes a good reference because it touches AI Providers, Connections, database settings, and branding.


Prerequisites

Before installing the Chat App, ensure the following are in place:

  1. AI Provider configured - At least one active provider with a chat model (e.g. OpenAI gpt-4o, Anthropic claude-sonnet-4-5). See AI Provider.
  2. Jina connection configured (optional, for web search) - A connection of type jina with a valid API key. See Jina AI.
  3. Your account has the App Manager role.

Installation

  1. Go to App-Store in the sidebar.
  2. Find core-operaide-chat in the card grid. It is a Core App and available on every fresh instance.
  3. Click Deploy at the bottom of the card.

Configuration

The wizard opens with the Chat App pre-selected. Enter an instance name (e.g. chat) and proceed to the settings step. The Chat App registers multiple Reaktors, each with its own settings.

Chat Reaktor

SettingDescriptionDefault
System PromptInstructions sent to the AI model at the start of every conversationA general-purpose assistant prompt (in German by default - adjust as needed)
Jina Connection NameName of the Jina connection used for web search. Must match a configured connectionjina

UI Configuration Reaktor

SettingDescriptionDefault
App TitleThe name displayed in the chat interfaceOperaide
Accent ColorHex color code for UI highlights (e.g. #10a37f)#10a37f
Brand Logo (Dark Mode)Logo image URL or base64 data URI (recommended: 64x64 px square PNG with transparency)(empty)
Brand Logo (Light Mode)Same as above, optimized for light backgrounds(empty)
Available ModelsJSON array defining which AI models users can choose from in the chat UI[{"provider":"openai","model":"gpt-4o","description":"GPT-4o"}]

The Available Models setting controls which provider/model combinations appear in the model selector. Each entry needs three fields:

[
{ "provider": "openai", "model": "gpt-4o", "description": "GPT-4o" },
{ "provider": "anthropic", "model": "claude-sonnet-4-5-20250514", "description": "Claude Sonnet 4.5" }
]

The provider and model values must match an active AI Provider configuration. The description is the label users see in the selector.

Database Reaktors

The Chat App uses a SQLite database for chat history, assistants, and templates. Multiple Reaktors (chat-api, message-api, assistant-api, template-api, template-folder-api, template-favorite-api) each require a Database ID setting.

To obtain the database ID, navigate to Agent DB in the sidebar. If no database exists yet, create one - give it a name (e.g. chat-db) and note the generated ID. Enter this same ID in every Reaktor that asks for a Database ID so they all share a single database. See Agent DB for details on creating and managing databases.

Template Wizard Reaktor

SettingDescriptionDefault
Template Generation System PromptSystem prompt used when generating prompt templates(built-in default)
Provider / ModelAI provider and model used for template generation(from AI Provider)

Verification

After deployment, open the Chat App and confirm:

  • Start a conversation - the AI should respond using the configured provider and model.
  • Try web search (if Jina is configured) - ask a question about current events and check that web results are included.
  • Upload a file - verify that PDF and text file handling works.
  • Check branding - the title, logo, and accent color should match your settings.