# amidabuddha/unichat-mcp-server [Health: Active]

**Category:** 🛠️ Other Tools and Integrations  
**Repository:** https://github.com/amidabuddha/unichat-mcp-server  
**GitHub Stars:** 40  
**Views:** 3  
**Installs:** 0  
**Upvotes:** 0  
**Directory Page:** https://allmcps.com/mcp/amidabuddha-unichat-mcp-server

## Description
/📇 ☁️ - Send requests to OpenAI, MistralAI, Anthropic, xAI, Google AI or DeepSeek using MCP protocol via tool or predefined prompts. Vendor API key required

## Tools
Capabilities this server exposes over MCP:

- **unichat** — Send a request to unichat

## Claude Desktop Quick Installation
Install path detected from listing signals. Uses `npx` (confidence: high):

```json
"mcpServers": {
  "unichat-mcp-server": {
    "command": "npx",
    "args": ["-y","@smithery/cli"],
    "env": {
      "UNICHAT_MODEL": "",
      "UNICHAT_API_KEY": "",
      "UNICHAT_BASE_URL": ""
    }
  }
}
```

**Requires environment variables:** `UNICHAT_MODEL`, `UNICHAT_API_KEY`, `UNICHAT_BASE_URL` — the values above are empty placeholders; fill in real credentials before running (see the repository for what each one is for).

## Documentation

## What amidabuddha/unichat-mcp-server MCP server does

The amidabuddha/unichat-mcp-server MCP server connects an MCP client to a selected language model through the Unichat configuration. It can send requests to OpenAI and Anthropic, as well as providers that expose an OpenAI-compatible API. The README identifies MistralAI, xAI, Google AI, DeepSeek, Alibaba, and Inception as examples of compatible providers.

The server is suited to clients that need model responses through MCP rather than a provider-specific client integration. A vendor API key is required, and the model is selected through an environment variable.

## How it works

The server implements one MCP tool named `unichat`. Its required input is a string called `messages`, and it returns the provider's response. The configured model and API credentials determine which backend receives the request.

Four predefined prompts cover common programming workflows:

- `code_review` accepts code and asks for best-practice, issue, and improvement feedback.
- `document_code` accepts code and generates documentation such as docstrings and comments.
- `explain_code` accepts code and requests a detailed explanation.
- `code_rework` accepts code plus optional requested changes and applies those changes.

For OpenAI-compatible providers, set `UNICHAT_BASE_URL` to the provider's compatible API endpoint. When this variable is present, the server accepts the configured model without validating it against Unichat's built-in model list.

## Setup and configuration

The amidabuddha/unichat-mcp-server MCP server can be launched as a published Python package with `uvx`:

```json
{
  "command": "uvx",
  "args": ["unichat-mcp-server"],
  "env": {
    "UNICHAT_MODEL": "SELECTED_UNICHAT_MODEL",
    "UNICHAT_API_KEY": "YOUR_UNICHAT_API_KEY"
  }
}
```

`UNICHAT_MODEL` selects the model, while `UNICHAT_API_KEY` supplies the vendor credential. For a compatible provider, add `UNICHAT_BASE_URL` with that provider's API endpoint. The README also documents running the project from a local source directory with `uv`, but that configuration requires replacing the example source path.

Claude Desktop configuration is documented for macOS and Windows. The README also links to a hosted deployment through Fronteir AI, although no endpoint details are provided in the supplied material.

## Tools and capabilities

The core capability of the amidabuddha/unichat-mcp-server MCP server is provider-agnostic text generation through the `unichat` tool. The predefined prompts make common code-oriented tasks available without constructing a prompt from scratch. The server does not list separate tools for each provider; provider selection is handled through environment configuration.

The implementation runs over MCP stdio in the documented local configurations. MCP Inspector is suggested for debugging stdio deployments, but it is a separate debugging utility rather than a capability of this server.

## Limitations and notes

A vendor API key is mandatory, so operating the server requires access to a configured model provider. The supplied material does not specify model pricing, request limits, streaming behavior, or support for provider-specific features. Compatibility with providers other than OpenAI and Anthropic depends on an OpenAI-compatible endpoint and the appropriate base URL.

The README names Claude Desktop as a supported configuration target. It does not provide setup instructions for other MCP clients in the supplied material. The project is released under the MIT license.

_Full upstream README: https://allmcps.com/mcp/amidabuddha-unichat-mcp-server/readme_