freestylefly/wesight

Open-source desktop AI agent workspace with one-click Claude Code, Codex, OpenClaw, Hermes Agent setup and custom LLM model routing.

What it solves

Terminal-native coding agents are powerful but often suffer from fragmented workflows. Users typically have to manage separate interfaces for terminal commands, model routing, permissions, file changes, and runtime metrics. WeSight consolidates these moving pieces into a single, unified desktop environment.

How it works

WeSight acts as a control console that installs, detects, and manages local agent CLIs (such as Claude Code, Codex, or DeepSeek-TUI) and provides a graphical user interface for them. It uses an Electron-based architecture to bridge terminal-based agents with a visual workspace. It manages model providers (like OpenAI, Anthropic, or Ollama), routes messages to IM channels like Feishu, and provides a live dashboard to monitor performance metrics like token usage and latency.

Who it’s for

  • Developers using terminal-based coding agents who want a visual workspace.
  • Users looking to manage multiple AI model providers and agent engines from one place.
  • People who want to automate tasks via scheduled agent jobs or custom skills.

Highlights

  • Unified Agent Engines: Supports Claude Code, Codex, Kimi Code, OpenClaw, Hermes Agent, OpenCode, Qwen Code, and DeepSeek-TUI.
  • Visual Workspace: Provides a live view of file writes, code changes, tool activity, and generated artifacts.
  • Runtime Dashboard: Tracks engine performance, including TTFT, TPS, token usage, and cost.
  • SkillHub Marketplace: Allows users to discover and install local skills for research, automation, and document processing.
  • Desktop Companion: Features a lightweight desktop pet and studio view to monitor active agent tasks.