Skip to main content
Glama
kent013

mcp-openai-extended

by kent013
OverviewSchemaSourceRelated ServersDiscussionsAdmin
JavaScript
Remote

MCP OpenAI Extended

Extended Model Context Protocol (MCP) server for OpenAI with GPT-5 family support, including specialized Codex models.

Features

  • ✅ GPT-5 family support (gpt-5, gpt-5.1, gpt-5.2, chat-latest/pro/mini/nano/codex variants)

  • ✅ GPT-4o and o1 series support

  • ✅ Responses API support for Codex models (*codex*)

  • ✅ Automatic API endpoint selection (Responses API for codex, Chat Completions for others)

  • ✅ Works with Claude Desktop, Claude Code, and any MCP-compatible client

Related MCP server: OpenAI API MCP Server

Supported Models

GPT-5 Family

  • gpt-5-codex - Specialized for code and design reviews (default, uses Responses API)

  • gpt-5.3-codex

  • gpt-5.3-codex-spark

  • gpt-5.2-chat-latest

  • gpt-5.2-pro

  • gpt-5.2

  • gpt-5.2-codex

  • gpt-5.1-chat-latest

  • gpt-5.1-codex

  • gpt-5.1-codex-mini

  • gpt-5.1-codex-max

  • gpt-5.1

  • gpt-5-chat-latest

  • codex-mini-latest

  • gpt-5-pro

  • gpt-5

  • gpt-5-mini

  • gpt-5-nano

Note: Newer Codex model IDs may require account-level access before they work in API requests.

GPT-4 Series

  • gpt-4o - GPT-4 Omni

  • gpt-4o-mini - Smaller GPT-4o

o1 Series

  • o1 - Reasoning model

  • o1-preview - Reasoning model preview

  • o1-mini - Smaller reasoning model

Installation

NPM (Recommended)

npm install -g @kent013/mcp-openai-extended

Via npx (GitHub)

npx github:kent013/mcp-openai-extended

Local Development

git clone https://github.com/kent013/mcp-openai-extended.git
cd mcp-openai-extended
npm install
npm run build

Configuration

⚠️ Security Warning

NEVER commit your API key to version control!

  • Keep your OPENAI_API_KEY in environment variables or secure configuration files

  • Add configuration files containing API keys to .gitignore

  • If you accidentally commit an API key, revoke it immediately at https://platform.openai.com/api-keys

  • Rotate your API keys regularly

Claude Desktop / Claude Code

Add to your .mcp.json or claude_desktop_config.json:

{
  "mcpServers": {
    "openai": {
      "command": "npx",
      "args": ["-y", "@kent013/mcp-openai-extended"],
      "env": {
        "OPENAI_API_KEY": "your-openai-api-key-here"
      }
    }
  }
}

Local Development

{
  "mcpServers": {
    "openai": {
      "command": "node",
      "args": ["/path/to/mcp-openai-extended/dist/index.js"],
      "env": {
        "OPENAI_API_KEY": "your-openai-api-key-here"
      }
    }
  }
}

Usage

The server provides a single tool: openai_chat

Default model is gpt-5-codex (automatically uses Responses API).

Example: Design Review with GPT-5.2 Codex (Default)

{
  "name": "openai_chat",
  "arguments": {
    "messages": [
      {
        "role": "system",
        "content": "You are an experienced system architect."
      },
      {
        "role": "user",
        "content": "Review this code:\n\n```php\n$user = User::find($id);\nMail::to($user->email)->send(new TestMail());\n```"
      }
    ]
    // model defaults to "gpt-5-codex"
  }
}

Example: General Reasoning with GPT-5 Chat Latest

{
  "name": "openai_chat",
  "arguments": {
    "messages": [
      {
        "role": "user",
        "content": "Explain quantum computing in simple terms"
      }
    ],
    "model": "gpt-5-chat-latest"
  }
}

API Endpoints

This server automatically selects the appropriate OpenAI API endpoint:

  • Responses API (/v1/responses): For Codex models (*codex*)

  • Chat Completions API (/v1/chat/completions): For all other models (gpt-4o, gpt-5-chat-latest, etc.)

Environment Variables

  • OPENAI_API_KEY (required): Your OpenAI API key

Getting Your API Key

  1. Visit https://platform.openai.com/api-keys

  2. Create a new API key

  3. Store it securely - never share or commit it to version control

  4. Set appropriate usage limits to prevent unexpected charges

Best Practices

  • Use environment-specific API keys (development, staging, production)

  • Enable usage alerts in your OpenAI account

  • Regularly review API usage and rotate keys

  • Use OpenAI's API key restrictions feature to limit key permissions

Development

# Install dependencies
npm install

# Build
npm run build

# Run locally
node dist/index.js

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

License

MIT

Credits

Based on mcp-openai by mzxrai.

Extended to support GPT-5 family and specialized models.

Available Tools

1 tool
openai_chatC

Send messages to OpenAI using a specified model. Supports GPT-4o, o1, and GPT-5 series models.

ParametersJSON Schema
NameRequiredDescriptionDefault
modelNoModel to use for completion (gpt-4o, gpt-4o-mini, o1, o1-preview, o1-mini, gpt-5/gpt-5.1/gpt-5.2 families, and codex variants)gpt-5-codex
messagesYesArray of messages to send to the API

TDQS

C2.9/5.0
Behavior2/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

No annotations are provided, so the description carries the full behavioral burden, yet it says nothing about cost, latency, rate limits, token/context limits, auth requirements, or whether calls are billed or stateful. It only names supported model families, which is closer to schema duplication than behavioral disclosure.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Conciseness4/5

Is the description appropriately sized, front-loaded, and free of redundancy?

Two short sentences, front-loaded with the core action; nothing is padded. The second sentence is mildly redundant with the enum in the schema but is not wasteful enough to penalize further.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness3/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

For a simple two-parameter call with fully covered schema and no output schema, the description is minimally adequate. However, with zero annotations it should at least flag cost/rate-limit behavior or model-selection tradeoffs, which it omits.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters3/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

Schema description coverage is 100%, so both parameters (model enum and messages array with role/content) are already documented, establishing a baseline of 3. The description's mention of 'GPT-4o, o1, and GPT-5 series' merely restates a subset of the enum and adds no format, default, or selection semantics.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose4/5

Does the description clearly state what the tool does and how it differs from similar tools?

States a clear verb (send) and resource (messages to OpenAI) with the model-selection dimension named. It distinguishes itself from any generic LLM-call sibling by naming the vendor, but there are no siblings to differentiate against and it doesn't clarify what the tool returns.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines2/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

No guidance on when to use this tool, when not to, or how to choose among the listed model families (e.g. cheap vs reasoning vs codex variants). The reader must infer usage entirely.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.

Tool Schema Changelog

Recent tool additions, removals, and schema changes observed during successful MCP inspections.

  1. 1 tool updatev1.2.0
    • First observedopenai_chat
Deploy Server
One endpoint for every MCP toolRayrun runs every MCP server behind one URL and holds the credentials. Each call is decided per client, tool, and argument, then recorded.ray.runAd

TDQS

B3.2/5.0

Scored across 1 tool

Disambiguation5/5

With only a single tool, there is no possibility of misselection or overlap. The one tool's purpose is unambiguous.

Naming Consistency4/5

openai_chat uses a clear, readable snake_case verb_noun convention that fits the provider prefix. With only one name there is no pattern to validate consistency against, so it cannot earn full marks.

Tool Count3/5

A single tool is thin for a server explicitly branded as 'extended' with multi-model support. The count is defensible for a minimal chat wrapper but feels under-scoped for the stated purpose.

Completeness3/5

Core chat completion is covered, but obvious companions are missing: model listing/discovery, conversation or state management, and other OpenAI surfaces implied by 'extended'. Agents can do basic work but hit dead ends beyond a single completion call.

Maintenance

ActivityInactive
ResponsivenessNo issues

Resources

Related MCP Connectors

Related MCP Servers

  • Customized MCP Server

    MorvanZhou
    A
    license
    Not graded
    quality
    D
    maintenance
    A server that integrates the MCP library with OpenAI's API, allowing users to interact with various tools, such as the weather tool, through natural language queries.
    MIT

  • SwiftOpenAI MCP Server

    jamesrochabrun
    A
    license
    Not graded
    quality
    C
    maintenance
    A universal server that enables MCP-compatible clients (like Claude Desktop, Cursor, VS Code) to access OpenAI's APIs for chat completions, image generation, embeddings, and model listing through a standardized interface.
    5 npm
    7
    MIT