Skip to content
English
  • There are no suggestions because the search field is empty.

Basic User Guide for LibreChat

LibreChat – Basic User Guide

With LibreChat, you can easily collaborate with chat agent Wilma and create your own workflows. This guide will take you step by step through the most important functions of LibreChat.

  1. User Registration
    1. Create a New User
    2. Login with Single Sign-On (SSO)
  2. Navigation
    1. Collapse and Expand Taskbar
    2. Start a New Chat
    3. View Chat History
  3. Chat Agents
    1. Create and Edit Chat Agents
    2. Chat Agents Profile: Adjust System Instructions
    3. Upload File Context
    4. Enable Artifacts
    5. Choose an AI Model
    6. Select MCP Server Tool
    7. Select Chat Agents
  4. Prompts Overview
    1. Create, Edite and Delete Prompts
    2. Share Prompts with Everyone
    3. Tips and Notes Creating Prompts
    4. Using Variables in a Prompt
  5. Error Analysis
    1. Download Chat Log

       

      1. User Registration

      Create a New User

      To register a new user in Wilma, open your Wilma instance in your browser and click on "Sign up" at the login screen. Enter your business email address and choose a secure password, then click "Continue" to create your account. Once registration is complete, you can log in with your new credentials.

        


       

      Login with Single Sign-On (SSO)

      For enterprise users, Single Sign-On (SSO) with Microsoft/Azure Entra is recommended or may be required by your organization’s security policies. Please check with your IT admin if SSO needs to be set up.

        



      2. Navigation

      Collapse and Expand Taskbar

      To give you more space for your chats, you can collapse or expand the taskbar on the left edge. Simply click the menu icon at the top left. This way, you can flexibly adjust the view to your needs.

      Wilma-sidebar2

       


       

      Start a New Chat

      To start a new chat, click the “New Chat” icon, usually displayed as a plus sign. You’ll immediately receive a new chat window where you can enter your request. This allows you to conduct multiple conversations with different agents in parallel.

      Wilma-newchat


       

      View Chat History

      Your entire chat history is automatically saved and clearly listed in the taskbar. You can access previous conversations at any time and continue directly in an existing conversation. This makes it easier to track requests and discussions.

      Wilma-history

      Each chat will be visible only to the user who initiated it. Other users will not have access to chat histories from other users.

       

      3. Chat Agents

      Create and Edit Chat Agents

      You can create your own chat agents or modify existing agents. To do so, select the “Chat Agents” area in the taskbar and click “Create Agent” or select an existing agent to edit. This allows you to design agents according to your requirements.

      Wilma-createnewagent2


       

      Chat Agents Profile: Adjust System Instructions

      Each agent can be equipped with its own instructions/system instructions, which controls the agent’s behavior. In the editor, you enter instructions on how the agent should respond to messages. This allows you to customize your agent’s responses and personality.

      System instructions

      For each zapliance Agent, your Customer Success contact person will provide you with the specific system instruction. 



      Upload File Context

      In the agent editor, upload a “File Context” so that the agent can access the information contained in the file. If a file is uploaded here, the agent will primarily use the information from this file for its analysis. If no file is provided, the agent is free to obtain the information from other sources. 

      FileUpload


       

      Enable Artifacts

      To allow the Chat Agent to include tables, lists, or other formatted content in its responses, you need to enable this feature. 

      artifacts2


       

      Choose an AI Model

      In chat agent settings, you’ll find the "Set Model" option, which allows you to choose your desired AI model. Your zapliance Customer Success Manager will instruct you during the onboarding process on which model and parameters (e.g., Temperature) should be used to achieve the best results for your project. 

      AIModel2


       

      Select MCP Server Tool

      To connect your project data in LibreChat, go to the agent settings and select your desired MCP server tool. This will allow the chat agent to communicate with the project data of your other zapliance Agents (Wolfgang, Werner, or Wickie). If you do not see any available MCP server tool after running zapAnalytics, please contact your zapliance Customer Success Manager.


       

      Select Chat Agents

      Before starting a new chat, you can select the appropriate agent from a list. Each agent specializes in certain topics, tasks or even date scope. This ensures you always chat with the right agent.

      SelectAgent


       

      4. Prompts Overview

      Create, Edit and Delete Prompts

      In each agent, you can store your own prompts—predefined questions or instructions. Go to the “Prompts” section in the agent editor and create a new prompt. This saves you time and improves the quality of responses.

      Wilma-createnewprompt

      To previewedit or delete an existing prompt, select the "..." menu.

      Wilma-editprompt

      For every zapliance product, your Customer Success Manager will provide a standard prompt library with predefined text relevant to your use case.

       

      Share Prompts with Everyone

      You have the option to share your created prompts with other team members. To do this, edit your prompt and click on the sharing icon at the top right. Next, enable the option to share with everyone and set the appropriate permission level. This way, all team members can benefit from helpful templates and best practices. 

      SharingPrompt


       

      Tips and Notes Creating Prompts

      Make sure to phrase prompts clearly and unambiguously so the chat agent can help you precisely. Try different formulations to achieve the best results. Use short, concise sentences and avoid ambiguous terms.

      Best Practice Prompt Example: Please create a table listing all indicators with hits for company code 2000 in fiscal year 2023. The table should include the following columns: Indicator ID, Indicator Name, Number of Hits, Relevance (yes/no), and a brief justification for the relevance assessment. Only consider manual and semi-automated postings, and exclude technical or system-generated entries. If possible, sort the table by the number of hits in descending order.


       

      Using Variables in a Prompt

      When creating a prompt in the prompt library, you can define variables by placing them in double curly brackets, like . Each time you use the prompt, you will be asked to fill in these variables. Using variables allows you to reuse a single prompt for different parameters, so you don’t need to save a separate prompt for each analysis.


       

      5. Error Analysis

      Download Chat Log

      If an error occurs, you can download chat logs and SQL statements. These help you or the support team to quickly analyze and resolve problems. Simply click the download icon in the editor area and select the type "text.txt".

      Wilma-log


       

      Help & Support

      If you have any questions about any of the steps or need assistance, please contact the zapliance team.