Communication Category Plugins in the OpenAI Repository: 11 Examples with Implementation Details

The OpenAI plugins repository identifies Communication category plugins through a "category": "Communication" field in each plugin's plugin.json manifest, including popular integrations like Zoom, Slack, Gmail, Microsoft Teams, and Otter AI that enable messaging, video conferencing, and email workflows.

The openai/plugins repository hosts reference implementations for extending OpenAI models through external APIs. Each plugin declares its functional domain via a structured metadata file, with Communication category plugins specifically handling real-time messaging, video conferencing, email management, and meeting transcription. These implementations reside in dedicated subdirectories under plugins/, with their classification explicitly defined in <plugin-root>/.codex-plugin/plugin.json.

How Communication Category Classification Works

The Plugin Manifest Structure

Every plugin in the repository defines its capabilities in a plugin.json file located at .codex-plugin/plugin.json within its root directory. The category field serves as the primary classifier, with Communication plugins explicitly setting "category": "Communication". This metadata allows the OpenAI platform to filter and route function calls to appropriate services based on user intent.

OpenAPI Specification Integration

Each manifest file exposes an OpenAPI specification that describes available endpoints, request schemas, and authentication methods. According to the openai/plugins source code, the platform uses these specifications to construct valid API calls when the model determines that Communication functionality is required.

Notable Communication Category Plugins

Video Conferencing Solutions

• Zoom: Implements the Zoom App SDK, Meeting SDK, and Video SDK for in-app video, chat, and app-to-app messaging. Configuration resides in plugins/zoom/.codex-plugin/plugin.json.
• Microsoft Teams: Exposes Teams-specific actions including message sending, meeting creation, and channel data retrieval. Manifest location: plugins/teams/.codex-plugin/plugin.json.

Email Management Platforms

• Gmail: Provides full-featured integration for sending, searching, and labeling emails via Google's API. Defined in plugins/gmail/.codex-plugin/plugin.json.
• Outlook Email: Interfaces with Microsoft Graph to send, read, and organize Outlook emails. Manifest: plugins/outlook-email/.codex-plugin/plugin.json.
• Superhuman: Enhances email workflows with thread-level communication timelines and smart reply capabilities. Located at plugins/superhuman/.codex-plugin/plugin.json.

Team Messaging and Chat

• Slack: Enables posting, reading, and reacting to Slack messages alongside channel management functions. Manifest: plugins/slack/.codex-plugin/plugin.json.
• Fyxer: Offers secure messaging and file-sharing capabilities for team collaboration. Configuration: plugins/fyxer/.codex-plugin/plugin.json.
• Granola: Implements a lightweight chat-bot framework designed for synchronous messaging scenarios. Found in plugins/granola/.codex-plugin/plugin.json.

Meeting Intelligence and Transcription

• Otter AI: Captures live audio, transcribes conversations in real-time, and shares communication notes. Manifest: plugins/otter-ai/.codex-plugin/plugin.json.
• Fireflies: Automates meeting transcription and generates follow-up communication summaries. Located at plugins/fireflies/.codex-plugin/plugin.json.
• Read AI: Summarizes meeting transcripts and extracts communication-focused insights for briefs. Configuration: plugins/read-ai/.codex-plugin/plugin.json.

Technical Implementation and Integration

Python Integration Example

The following demonstrates invoking the Gmail plugin through the OpenAI Chat Completion API:

import openai

response = openai.ChatCompletion.create(
    model="gpt-4-plugins",
    messages=[
        {"role": "user", "content": "Send an email to [email protected] saying \"Meeting at 3 PM\"."}
    ],
    plugins=[
        {
            "type": "plugin",
            "id": "gmail",
            "manifest_url": "https://github.com/openai/plugins/blob/main/plugins/gmail/.codex-plugin/plugin.json"
        }
    ]
)

print(response["choices"][0]["message"]["content"])

The plugins parameter references the manifest URL defined in plugins/gmail/.codex-plugin/plugin.json, triggering the plugin's send-email operation when the model determines email functionality is required.

Manifest Schema Requirements

Each Communication plugin must declare its category explicitly to be recognized by the platform. The plugin.json structure includes fields for plugin ID, description, authentication details, and the OpenAPI spec URL. As implemented in plugins/slack/.codex-plugin/plugin.json and similar paths, this standardized format ensures consistent integration across video, email, and messaging platforms.

Summary

• Communication category plugins in the openai/plugins repository are identified by the "category": "Communication" field in their plugin.json manifests.
• Notable examples include Zoom, Microsoft Teams, Slack, Gmail, Outlook Email, Superhuman, Otter AI, Fireflies, Read AI, Fyxer, and Granola.
• Each plugin stores its metadata in .codex-plugin/plugin.json within its respective plugins/<name>/ directory.
• Integration requires referencing the plugin's manifest URL and ID in the Chat Completion API's plugins parameter.
• These plugins expose OpenAPI specifications that enable video conferencing, email management, team messaging, and meeting transcription capabilities.

Frequently Asked Questions

How does the OpenAI plugins repository classify Communication plugins?

The repository uses a category field inside each plugin's plugin.json file. When this field contains the value "Communication", the plugin is categorized as handling messaging, email, video calls, or real-time interaction capabilities. This classification metadata resides in <plugin-root>/.codex-plugin/plugin.json.

What types of functionality do Communication category plugins provide?

Communication plugins cover four primary domains: video conferencing (Zoom, Microsoft Teams), email management (Gmail, Outlook, Superhuman), team messaging (Slack, Fyxer, Granola), and meeting intelligence (Otter AI, Fireflies, Read AI). Each exposes specific endpoints through OpenAPI specifications defined in their manifest files.

How do you invoke a Communication plugin through the OpenAI API?

You include a plugins array in your ChatCompletion.create request, specifying the plugin's id (folder name) and manifest_url pointing to the raw plugin.json file. The model then uses the OpenAPI schema defined in plugins/<name>/.codex-plugin/plugin.json to construct appropriate API calls to the external service.

Where are the manifest files located for Communication plugins?

Each plugin stores its manifest at plugins/<plugin-name>/.codex-plugin/plugin.json within the openai/plugins repository. For example, Slack uses plugins/slack/.codex-plugin/plugin.json, while Zoom's configuration resides at plugins/zoom/.codex-plugin/plugin.json.

{
  "mcpServers": {
    "instagit": {
      "command": "npx",
      "args": ["-y", "instagit@latest"]
    }
  }
}