// the find
iOfficeAI/AionUi
Free, local, open-source 24/7 Cowork app for OpenClaw, Hermes Agent, Claude Code, Codex, OpenCode, Gemini CLI and 20+ more CLI | Customize your assistants | Star if you like it!
AionUi is an Electron desktop app that puts a unified GUI around AI CLI agents (Claude Code, Codex, Gemini CLI, etc.) while also shipping its own built-in agent with file ops, web search, MCP tool support, and cron scheduling. It targets developers and knowledge workers who want a persistent, cross-platform workspace for running AI agents unattended. The 28K stars arrived fast, largely via viral YouTube coverage.
The extension SDK is genuinely well-architected — JSON manifests, contributed assets, separate skill/assistant/theme layers, with working examples in the repo. The MCP unified management solves a real pain point: configure MCP servers once and they sync across all detected agents rather than re-configuring Claude Code, Codex, etc. separately. Cron scheduling with per-task model, workspace directory, and reasoning effort settings is a concrete feature, not marketing — unattended agent execution with missed-trigger detection on wake is useful. The auto-detection of installed CLI tools, rather than requiring manual wiring, is a good UX decision that respects what developers already have installed.
The 'zero setup, no configuration' pitch collapses immediately — you still need an API key, and the built-in agent is only as capable as the model behind that key; there is no free compute here. YOLO/Full-Auto mode (bypass all permission prompts) combined with cron scheduling and full file-system access is a serious footgun — an agent running unattended with no confirmation dialogs on a cron schedule is how you lose files. The README's comparison table benchmarks against 'Claude Cowork' at $100/month, which is not a real Anthropic product — it's a strawman lifted from Claude's subscription pricing applied to a feature that doesn't exist, making the whole comparison misleading. WeChat personal account integration is a ToS violation risk that could get users' accounts banned without warning.