LibreChat

LibreChat is an open-source chat application for different LLM models. Using Anyquery, you can connect LibreChat to your local and online applications (such as Spotify, GitHub, Discord, etc.) and interact with them using natural language.

Prerequisites

Before you begin, ensure that you have the following:

• A working installation of Anyquery (v0.4.0). Follow the installation guide if you haven’t installed it yet.
• A few integrations set up in Anyquery. Refer to each integration guide for more information.
• LibreChat hosted somewhere.

Connect LibreChat to Anyquery

To connect Anyquery to LibreChat, you can either use the HTTP api (recommended) or the model context protocol.

Ensuring agents are enabled

For both protocols, you need to enable Agents. By default, they should be enabled, and available in the Agents section of the model selection screen. If not, ensure your librechat.yaml configuration file looks like this:

version: 1.2.1

interface:
  agents: true

endpoints:
  agents:
    recursionLimit: 25 # Can be increased if needed
    disableBuilder: false
    capabilities: ["actions", "tools"] # You can add other capabilities here
  ...

If you still cannot see the Agents section, check your environment variables. The ENDPOINTS variable should include agents.

ENDPOINTS=azureOpenAI,gptPlugins,agents,openAI

HTTP API

1. Start Anyquery

   In a terminal, run the following command to start Anyquery:

   Terminal

   anyquery gpt

   This command will start a tunnel to Anyquery and provide you a bearer token. You must then use this token to authenticate your requests to Anyquery.

   Terminal

   julien@MacBook-Air-Julien anyquery % anyquery gpt
   
   
   Your Anyquery ID is ikabcjde. This is your bearer token that you must set in chatgpt.com
   
   
   2025-02-10 21:29:12.293 [I] [client/service.go:295] try to connect to server...
   2025-02-10 21:29:12.453 [I] [client/service.go:287] [cc07e04c16c1e59f] login to server success, get run id [cc07e04c16c1e59f]
   2025-02-10 21:29:12.454 [I] [proxy/proxy_manager.go:173] [cc07e04c16c1e59f] proxy added: [ikabcjde]
   2025-02-10 21:29:12.494 [I] [client/control.go:168] [cc07e04c16c1e59f] [ikabcjde] start proxy success

2. Set up LibreChat

   Select the Agents section in the model selection screen, and create a new one.

   

   In the right panel, fill in the fields for your new agent.

   • Name and Description: Give your agent a name and a description.
   • Instructions Go to this gist and copy the content to the instructions field.
   • Model: Select the LLM model you want to use with LibreChat.

   Click on Create to save your agent.

   

   Now that your agent is created, click on Add actions to connect it to Anyquery.
   Under authentication, select API Key, select Bearer as Auth Type, and paste the bearer token you received from Anyquery.

   Then, go to this gist and copy the content to the Schema field.

   
   

   Finally, click on Create to save your action. Go back, and then click on Save to save your agent.

   
   

   Congratulations 🎉. LibreChat is connected to Anyquery.

Model Context Protocol

Anyquery can also connect to LibreChat using the Model Context Protocol (MCP).

1. Set up LibreChat

   Open the librechat.yaml configuration file, and add the following configuration:

   version: 1.2.1
   
   mcpServers:
       anyquery:
           command: anyquery
           args:
               - mcp
   
   # Rest of your configuration

   Now, we will create a new agent in LibreChat. Select the Agents section in the model selection screen, and create a new one.

   

   In the right panel, fill in the fields for your new agent.

   • Name and Description: Give your agent a name and a description.
   • Instructions You can leave it empty.
   • Model: Select the LLM model you want to use with LibreChat.

   Now, click on Add Tools to add Anyquery as a tool. Go to the last page, and add:

   • listTables
   • describeTable
   • executeQuery

   

   Finally, click on Create to save your agent. Go back, and then click on Save to save your agent.

Selecting the agent

Now that you have set up your agent, go back to the Agents section in LibreChat, and select the agent you just created in the right panel

You can now start interacting with LibreChat using natural language. Here are a few examples:

• What can you do?
• Recommend me a song according to my history.
• Take this PDF, and open the references in my browser.
• Here is a link to my schedule, when should I schedule a meeting?