chat-with-openai
Relays a question to a configured AI Chat Provider.
LABSCO SUMMARY
Any chat API becomes a tool your assistant can call.
One small server covers every provider that speaks the OpenAI shape, and the configuration is four environment variables - so adding a second opinion from Perplexity or Groq is a config block rather than an integration.
What it is
A small TypeScript server that relays a question to any chat completions API that speaks the OpenAI SDK's shape - OpenAI, Perplexity, Groq, xAI, PyroPrompts and others. It exposes a single chat tool, named after the provider you configure, so several providers can sit side by side as separate entries.
What you get
- A question relayed to any OpenAI-SDK-compatible chat completions endpoint, and the answer returned in your conversation
- The provider chosen entirely in configuration - key, display name, model and base URL - with no code change
- The tool named after the provider you configured, so a client with three of these set up shows three distinct tools
- Several providers at once: add the same server again with a different set of environment variables
Requirements
Node, with the package run through npx or built from a clone. Four environment variables per provider: AI_CHAT_KEY, AI_CHAT_NAME, AI_CHAT_MODEL and AI_CHAT_BASE_URL. Each provider you want is its own entry in the client config.
Setup effort
Add the configuration to your client
USE CASES
01
Asking a second model the same question
Hand the question over inside the conversation and read both answers side by side.
02
Routing a search-flavoured question to a provider that is good at it
Configure Perplexity as its own entry, so web-shaped questions can be passed to it by name.
03
Trying a different model without changing anything else
Point the model and base URL variables at the candidate, and the same tool now talks to it.
04
Keeping several providers available at once
Add the server once per provider, each with its own display name, so your client lists one tool per provider.
1 tool
Counted once with only startup setup. Other versions, modes or optional credentials may expose a different list.
BEFORE YOU INSTALL
What to know first
Tools you get
1 tool when we connected. The README calls this tool chat, but the server registers it as chat-with- plus the AI_CHAT_NAME you set, lower-cased: chat-with-openai for the README's OpenAI example. That is the name your client shows.
Requires
- npx on PATH
- AI_CHAT_KEY in your environment
INSTALLATION
Connect your client
MCP Clients
Claude Desktop
CLAUDE DESKTOP · JSON
{
"mcpServers": {
"chat-openai": {
"command": "npx",
"args": [
"@pyroprompts/any-chat-completions-mcp"
],
"env": {
"AI_CHAT_KEY": "OPENAI_KEY",
"AI_CHAT_NAME": "OpenAI",
"AI_CHAT_MODEL": "gpt-4o",
"AI_CHAT_BASE_URL": "https://api.openai.com/v1"
}
}
}
}LibreChat
LIBRECHAT · YAML
mcpServers:
chat-perplexity:
type: stdio
command: npx
args:
- -y
- "@pyroprompts/any-chat-completions-mcp"
env:
AI_CHAT_KEY: pplx-012345679
AI_CHAT_NAME: Perplexity
AI_CHAT_MODEL: sonar
AI_CHAT_BASE_URL: https://api.perplexity.ai
PATH: /usr/local/bin:/usr/bin:/bin
