ShioriCode

Product Introduction

1. Definition: ShioriCode is a source-available, cross-platform desktop application (macOS, Windows, Linux) that functions as a unified graphical interface and workspace manager for multiple AI-powered coding agents, including Codex, Claude Code, Cursor, Gemini, Kimi, and a hosted Shiori provider.
2. Core Value Proposition: It exists to solve the fragmentation and context-loss problem in AI-assisted development by providing a single, persistent, and checkpointed desktop workspace where developers can run long-running, project-aware coding sessions across different AI models and seamlessly track, review, and integrate their output.

Main Features

1. Multi-Provider, Multi-Model Workspace: ShioriCode integrates several leading AI coding agents into one interface. This allows developers to compare outputs from models like Anthropic's Claude, OpenAI's Codex, Google's Gemini, and Cursor's agent side-by-side within the same project context, without switching between different CLIs or web interfaces.
2. Checkpointed, Thread-Based Sessions: The core of ShioriCode is its persistent session model. Instead of isolated prompts, it creates long-running, stateful "threads" that are aware of your entire code repository. Each interaction turn is automatically checkpointed using Git, allowing you to rewind to any previous state, review the agent's incremental changes, and maintain context over extended development tasks.
3. Real-Time Activity Streaming & Diff Review: All agent activity is streamed live to a centralized, readable timeline. This includes every file edit, shell command execution, and the agent's internal reasoning steps. Crucially, the app provides an integrated diff viewer to inspect all code changes generated in a turn, enabling approval, modification, or rollback without leaving the application.

Problems Solved

1. Pain Point: Context loss and workflow fragmentation when using multiple AI coding assistants. Developers often juggle separate terminals, web UIs, and prompts, losing the thread of complex tasks and struggling to track the cumulative changes made by an agent over time.
2. Target Audience: Professional software engineers, full-stack developers, and technical leads working on complex refactoring, feature implementation, or bug fixes that exceed the scope of a single AI prompt. It's particularly valuable for developers who actively evaluate and utilize multiple AI models (e.g., Codex for speed, Claude for reasoning) in their workflow.
3. Use Cases: Implementing a new API endpoint across a full-stack codebase; systematically refactoring a legacy module; debugging a complex, multi-file issue; conducting comparative analysis of how different AI models (Claude vs. Gemini) approach the same coding problem within your specific project.

Unique Advantages

1. Differentiation: Unlike single-provider IDE plugins (e.g., GitHub Copilot chat) or standalone CLI tools, ShioriCode is a model-agnostic desktop environment built specifically for orchestrating and observing long-running AI agent work. It prioritizes session persistence, multi-model comparison, and granular change audit over single-prompt completion.
2. Key Innovation: The integration of Git-based checkpointing at every agent turn transforms AI coding from a transactional prompt-response cycle into a version-controlled, stateful collaboration. This, combined with the unified activity stream, provides an unprecedented level of transparency and control over the AI-assisted development process.

Frequently Asked Questions (FAQ)

1. What is ShioriCode and how is it different from Cursor or GitHub Copilot? ShioriCode is not an AI model itself but a desktop workspace that runs agents like Cursor and Claude. While Cursor is an AI-centric IDE with its own built-in agent, ShioriCode lets you use Cursor's agent alongside others in a single, checkpointed session, focusing on managing multi-agent workflows and reviewing their history.
2. Is ShioriCode free to use? The ShioriCode desktop application is included as a feature for active paid subscribers of Shiori's Plus, Pro, or Max plans. It is not a standalone free product; access is tied to a Shiori subscription which provides the hosted Shiori agent and model access.
3. Does ShioriCode work with my local code repository? Yes, ShioriCode is designed to be project-aware. It reads your local Git repository to provide full context to the AI agents and uses Git internally to create checkpoints of every change made during an agent session, ensuring all work is traceable and revertible.
4. Can I use ShioriCode with OpenAI's GPT models or other AI providers? Currently, ShioriCode supports the listed providers: Codex (OpenAI), Claude (Anthropic), Gemini (Google), Cursor, Kimi, and Shiori's own hosted agent. It is built to be extensible, but direct integration with the standard ChatGPT API is not explicitly mentioned as a current feature.
5. What operating systems does ShioriCode support? ShioriCode is a cross-platform desktop application with official downloads available for macOS (Apple Silicon and Intel), Windows (x64), and Linux (via AppImage).