yohasebe/openai-chat-api-workflow

What it solves

This project provides a seamless way to integrate OpenAI's GPT models directly into the macOS Alfred 5 productivity app. It eliminates the need to switch between different browser tabs or applications to access AI capabilities, bringing text generation, image creation, and file analysis directly into the system's workflow.

How it works

The workflow connects to the OpenAI Chat API using a user-provided API key. It offers three primary interaction methods: direct commands within the Alfred UI, passing selected text via universal actions, and a locally hosted web interface for more complex interactions. It supports a wide range of OpenAI models, including those with reasoning capabilities, and handles various modalities including images, PDFs, and audio files.

Who it’s for

macOS users who use Alfred 5 and want to integrate AI-powered text, image, and voice tools into their daily productivity flow without leaving their current workspace.

Highlights

• Multimodal Capabilities: Supports image generation, image editing, file understanding (PDFs, Office docs, code), and speech-to-text/text-to-speech synthesis.
• Flexible Interfaces: Accessible via Alfred keywords, hotkeys, selected text, or a dedicated local web UI.
• Iterative Image Refinement: Allows users to refine generated images through follow-up prompts.
• Reasoning Control: Provides settings to adjust the "reasoning effort" for compatible GPT-5 series models to balance speed and quality.
• Privacy-focused: API calls are made directly between the workflow and OpenAI; the local web UI is constructed and run on the user's Mac.