Creating an agent in LibreChat

Questions

• How can I create an agent in LibreChat?
• What information can I add in an agent?
• How do I create an AI agent?
• How do I create an assistant in LibreChat?

Environment

• Access to LibreChat
• Current Bowdoin staff, faculty, and student

Answer

LibreChat is Bowdoin College's AI chat gateway platform, available at librechat.bowdoin.edu. One of its most powerful features is the Agent Builder, which allows you to create custom AI agents tailored to specific tasks or workflows.

An AI agent in LibreChat is a customized version of an AI model that you configure with a specific name, instructions, capabilities, and tools. Agents can be used to automate repetitive tasks, provide specialized assistance, search through documents, and much more.

This guide will walk you through the entire process of creating an agent.

What you'll need

• Access to LibreChat
• A clear idea of what you want your agent to do
• Any files or "artifacts" you want your agent to reference (optional)

Step 1: Open the Agent Builder

The Agent Builder is located in the right-hand side panel of the LibreChat interface. Here's how to access it:

• In the LibreChat chat window, locate the small Tab icon on the far right side of the interface (1). Click it to open the right-hand panel.
• Once the panel opens, you will see a dropdown menu with several options: Agent Builder, Prompts, Memories, Parameters, Attach Files, Bookmarks, and Hide Panel. Click Agent Builder (2) to begin creating your agent.
  
  
   
  Tip: The panel icon may look like a small sidebar or hamburger icon. If the panel is already open, look for the Agent Builder button in the drop-down at the top of the right panel.

Step 2: Fill in your agent's basic information

Once the Agent Builder is open, you'll see a Create New Agent form. This is where you define the core identity and behavior of your agent.

Name (Required)

Enter a clear, descriptive name for your agent. This name will appear in the interface when you select your agent for a conversation. Choose something that reflects the agent's purpose, such as "Research Assistant" or "Email Drafter".

Description (Optional)

Add a brief description of what your agent does. This is especially helpful if you plan to share the agent with others or if you have multiple agents and need to distinguish between them.

Category (Required)

Select a category for your agent from the dropdown menu. The default is "General," which works for most use cases. Categories help organize agents within the platform.

Instructions

This is the most important field — it defines how your agent will behave. Write clear, specific instructions that tell the agent:

• It's role and purpose (e.g. "You are a helpful research assistant for Bowdoin faculty")
• The tone and style to use (e.g. formal, casual, concise, polite)
• Any specific tasks it should focus on
• Any limitations or things it should avoid
• (Optional) You can also click the Variables button next to your instructions to add dynamic placeholders.

Model (Required)

Select the AI model that will power your agent. The default is GPT-5, which is suitable for most tasks. Different models have different strengths — for example, some are better at coding while others excel at creative writing.

Step 3: Configure Capabilities

After setting the basic information, scroll down to configure the Capabilities section. This is where you can give your agent access to additional features.

File Context

File Context allows you to upload reference documents that your agent will always have access to when responding. This is useful for:

• College or department policy documents
• Frequently referenced manuals or guides (such as PDFs of pages from this Support Portal!)
• Syllabi or course materials

Artifacts (Advanced Capability)

The Artifacts section lets you control whether your agent can generate structured outputs, such as charts, code blocks, or formatted documents. There are three toggles:

• Enable Artifacts - turns on the artifiacts feature for this agent
  • Toggle this on if you want the agent to produce formatted outputs
• Include shadcn/ui component instructions - for agents that generate UI components
  • Useful for people creating and building interfaces with React and shadcn/ui
• Custom Prompt Mode - Allows more advanced customization of how the agent responds.
  • Leave this off unless you know you have a specific need for it.

File Search (Advanced Capability)

File Search allows your agent to search through documents you upload. This is ideal when you want your agent to answer questions based on a specific set of documents (e.g., a knowledge base or policy library). Check the Enable File Search box to activate this feature, then upload your documents after the agent has been created.

Step 4: Add Tools, Actions, & Support Contact

Additional information that will be useful, especially if you share your agent.

Tools and Actions (Advanced Capability)

This section allows you to extend what your agent can do:

• Add Tools - connect pre-built tools to your agent, such as web search, calculators, or API connectors. Click "Add Tools" to browse available options.
• Add Actions - create custom actions that let your agent interact with external services via APIs. This is an advanced feature for users who want to connect the agent to external systems.

Support Contact

Fill in the Support Contact section if you want users to have a point of contact if they have questions about the agent:

• Name - the name of the person or team responsible for this agent
• Email - a contact email address (shown as a placeholder: support@example.com)

Advanced Settings

The Advanced button at the bottom of the form opens additional configuration options for power users. These settings are optional for most use cases and can be explored once you're comfortable with the basics.

Step 5: Create your agent

Once you have filled in all the required fields (Name, Category, and Model), you are ready to create your agent.

1. Review your settings
   • Scroll through the form one final time to make sure everything looks correct - especially the Instructions field.
2. Click the green Create button
   • At the very bottom of the form, click the large green "Create" button. Your agent will be created and saved immediately.
3. Upload files (if applicable)
   • After creation, return to the File Context or File Search sections and upload any relevant documents you want your agent to access.
4. Start chatting
   • Your new agent will now appear in LibreChat for you to use. Select it from the Model Selection drop-down to start a conversation and test how it works. (See Step 6. below)

Step 6: How to access My Agents

Once your agent has been created and saved, you can access it from the top-left corner, under the Model Selection drop-down where you typically select different models (e.g., OpenAI, Anthropic). All agents are organized under the “My Agents” category, which includes both agents you’ve created and those shared with you by others.

Tips for writing great agent instructions

The quality of your agent depends heavily on how well you write its instructions. Here are some best practices:

• Be specific about the agent's role. Instead of "You are a helpful assistant", try "You are a writing coach who helps Bowdoin students improve their academic essays."
• Set the tone. Specify whether responses should be formal, conversational, concise, or detailed.
• Define scope. Tell the agent what topics it should focus on and which to avoid.
• Use examples. Include sample interactions in the instructions to show the agent the style you expect.
• Iterate. Create the agent, test it, then refine the instructions based on the results

Troubleshooting

The Create button is grayed out

Make sure all required fields (marked with a red asterisk) are filled in: Name, Category, and Model are all required.

My agent isn't following the instructions

Try rewriting the Instructions field to be more direct and specific. Avoid vague language. Starting instructions with "You are..." and using clear directives tends to work well. You do not have to be polite, you wont' hurt the agent's feelings. The more direct, the better.

I can't upload files for File Context or File Search

Files can only be uploaded after the agent has been created. Create the agent first, then return to the agent's settings to upload documents.

I need help with advanced features

For questions about Tools, Actions, or Custom Prompt Mode, contact the Service Desk by submitting a request and we'll have someone reach out to help.

Additional Help

If you need further assistance, please contact the Bowdoin College Service Desk