<2-3 sentences summarising what the project addresses and solves, the target audience, and the benefit.> WeSight is an open-source desktop workspace designed to unify and simplify the management of local AI coding agents. It targets developers and power users by providing a single interface to install, configure, and interact with diverse agent engines like Claude Code, Codex, and Hermes Agent, alongside custom LLM routing and a visual environment for chat, tools, and file operations. The primary benefit is consolidating the complex, often fragmented, setup and operational aspects of terminal-native AI agents into an accessible desktop application.

How It Works

The application employs an Electron architecture with a React/Redux/Tailwind frontend, ensuring process isolation between the renderer and main Node.js processes via IPC bridges. This design allows WeSight to act as a central control console, detecting or installing local agent CLIs, managing unified model provider configurations (including OpenAI, Anthropic, Gemini, and custom endpoints), and routing tasks. Its advantage lies in abstracting the complexities of agent setup, model selection, permissions, and runtime monitoring into a cohesive graphical user interface.

Quick Start & Requirements

For macOS users, installation involves downloading a DMG file and dragging the application to the Applications folder. Preview builds may require manual quarantine attribute removal via xattr -cr /Applications/WeSight.app in the Terminal. Development requires Node.js (v24-v25) and npm. Initial releases prioritize macOS Apple Silicon. Official quick-start guides and the latest release are available via wesight.ai and the GitHub releases page.

Highlighted Details

• Broad Agent Engine Support: Integrates Claude Code, Codex, OpenClaw, Hermes Agent, OpenCode, Qwen Code, DeepSeek-TUI, and a built-in runtime.
• Unified Model Management: Centralizes configuration for official providers (OpenAI, Anthropic, Gemini) and numerous custom/local endpoints (Ollama, OpenRouter, etc.).
• IM Integration: Routes messages from channels like Feishu directly into supported agent workflows.
• Comprehensive Runtime Dashboard: Provides detailed metrics on engine/model usage, token counts, latency, cost, and agent step progression.
• Live Workspace & Skills: Features a real-time code editor, file diff viewer, task management, and a SkillHub marketplace for extending agent capabilities.
• Desktop Companion: Includes a visual "desktop pet" and studio view for active tasks.

Maintenance & Community

The project is sponsored by PPToken. Community engagement is primarily facilitated through a WeChat group (QR code valid until June 8, 2026) and a WeChat Official Account. Roadmap ideas are outlined within the repository.

Licensing & Compatibility

WeSight is released under the MIT license, permitting broad use, modification, and distribution, including for commercial purposes, without copyleft restrictions.

Limitations & Caveats

Initial releases are specifically targeted at macOS Apple Silicon, suggesting potential limitations or lack of support for other operating systems or hardware architectures. Preview builds are unsigned, requiring manual user intervention to bypass macOS Gatekeeper security warnings.