Operon is a unified desktop interface for AI coding agents. Use natural language to orchestrate multiple agents in one conversation ā for example, "plan with Codex, then implement with Opencode." Enjoy a CLI-consistent experience with multi-agent workflows, parallel sessions, and Telegram integration.
Multi-agent coding desktop, now on iPhone, Android and web
Give Claude Code, Codex & Gemini a real desktop and let them work on one repo together. New: iOS, Android & web clients ā the agent keeps running on your Mac, you steer it from your phone, end-to-end encrypted. Plus Computer Use & Chrome control. Open source.
OperonOne gateway. Many agents. Slack, Linear, Telegram.
Launched on May 14th, 2026
OperonDesktop GUI for ClaudeCode, Codex, Gemini CLI, Opencode
Launched on April 2nd, 2026
Reviews
No reviews yetBe the first to leave a review for Operon
Maker
š
Hi Product Hunt š
Operon is a desktop for AI coding agents that isn't tied to any one of them ā Claude Code, Codex, Gemini CLI, Copilot CLI, Cursor Agent, Kimi, Grok, OpenCode, or anything speaking ACP. It drives the official CLIs on your machine, so you use whichever subscriptions you already pay for. We don't resell model access, and your code never leaves your Mac.
We launched an earlier version here; since then the biggest change is that Operon is no longer a Mac-only app.
š One session, every device
⢠iOS (App Store), Android (APK) and a web client ā all reaching the same session that's running on your Mac
⢠Start a run at your desk, approve a diff from your phone, steer it from a browser ā the agent never stops
⢠Remote traffic is end-to-end encrypted between your approved device and your desktop; the relay only sees ciphertext
⢠Telegram and Slack bridges if you'd rather not install anything
š¤ Multiple agents, working together
⢠Multi-agent threads ā put two or more agents in one conversation, let them disagree, you decide
⢠Canvas workflows ā describe a pipeline in plain text; each node starts the moment its own inputs are ready, so branches truly run in parallel
⢠Isolated worktrees ā every dispatched task gets its own git worktree, so parallel agents never collide
⢠Spec-driven changes ā spec, plan and acceptance each get a human sign-off, and an independent verifier agent checks the result before you approve
š Also new since last time
⢠Computer Use ā agents can see and operate native macOS apps
⢠Chrome control ā drive your own Chrome through an extension, or the built-in browser
⢠Skills installable per project, not just globally
⢠Reworked long-term memory (local hybrid vector + keyword search)
⢠GitHub / Linear / Slack integrations ā agents pick up work from issues and PRs
Still there: per-turn diff review, live context & usage tracking, MCP servers, cron jobs, integrated terminal.
Free, open source (AGPL-3.0), macOS + iOS + Android + web. Would love to hear what you'd want from the mobile side ā that's where we're spending the most time right now.