Open-source Runner orchestrates Claude Code, Codex and other terminal agents in real TUIs, built with Rust+GPUI
jason3gb · reddit · 2026-11-19
A developer built Runner, a free, MIT-licensed open-source tool that lets multiple CLI coding agents work together in real terminals, with first-tier support for Claude Code, Codex, Antigravity CLI, Grok Build and pi.
Design: no ACP wrappers — every agent keeps its full native TUI; implemented natively in Rust with GPUI (Zed's UI framework) and alacrittyterminal, avoiding Electron. Each agent runs in a PTY managed by a background daemon (runnerd), so sessions survive app quits or crashes.
Key features
- Reusable crews, missions coordinated over a shared feed via the bundled runner CLI, with inline questions when agents need decisions
- Fixed turn-wasting polling: missions link under the launching chat, which gets one status line back on input/failure/interrupt/settle; best-effort mid-turn inbox delivery reaches active Claude Code, Codex, pi, Antigravity CLI and OpenCode v2 sessions
- git worktree support keeps parallel missions on separate branches; side panel offers Cmd-click file reading, Unified/Split diff views with base-branch selection, and read-only viewed-marking
- First launch detects installed CLIs and builds a starter Pair coding crew; themes sync live including Claude Code
Development process: the author only handles architecture, UI design in pen.dev, and diff review — agents wrote almost all the code; a coder/reviewer crew rewrote the frontend from Tauri+React to Rust+GPUI in about five days, and he now builds Runner inside Runner.
Platforms: native macOS (Apple Silicon/Intel) and Windows x64 without WSL, Linux in beta; an iOS TestFlight companion offers E2E-encrypted remote control with the relay server in invite-only beta for mainland China. Fully local — no Runner account, server, or proxy.
More from coding & agent
- Every on Claude Fable 5.2: Human Thinking at the Start Matters More as AI Builds More — every · 2026-11-19
- Semantic caching LLM answers by embedding similarity is unsafe — 138-pair test shows why — Narrow_Side_7778 · 2026-11-19
- Can ComfyUI refine existing mocap animations? A motion-to-motion workflow question — _Tito5_ · 2026-11-19
- Mycelium: an open-source framework that turns analytical project knowledge into a living dataset — arjunrajlab · 2026-11-19
- harness-hub open-sources RL-trained LoRA adapters tuned per agent harness, recipe coming Friday — yb2698 · 2026-11-19
- RL-Trained Hermes LoRA on Qwen3.8-27B Adds +13.7 Pts on SWE-Bench Pro While Using Fewer Tokens — yb2698 · 2026-11-19