{
  "id": 20935785,
  "title": "Run your coding agents in tmux and let a local model tell you which replies to read",
  "url": "https://urgent.news/2026/11/20/run-your-coding-agents-in-tmux-and-let-a-local-model-tell-you-which",
  "topic": "tech",
  "section": "Tech",
  "published": "2026-11-20T19:07:45.000Z",
  "source": {
    "name": "Dev.to",
    "slug": "dev-to",
    "url": "https://dev.to/voidcraft_world/run-your-coding-agents-in-tmux-and-let-a-local-model-tell-you-which-replies-to-read-2k30"
  },
  "original_language": "en",
  "account": "Two weeks ago, the author conducted 1,523 coding prompts, reviewing every outcome message. Dissatisfied, they integrated a local model, TLDRack, in front of the agents. Out of the initial prompts, 696 required further attention. TLDRack is an open-source tool (MIT license) compatible with macOS and Claude Code or Codex. This article outlines setting up and utilizing TLDRack. TLDRack operates your coding agents in parallel via tmux, displaying all sessions on a single local page. Sessions are named according to the prompt for easy identification. After a turn concludes, the local model assesses the final message, flagging it as read or requiring your review. Installation requires git, tmux, ttyd, Node.js 22.18 or later, uv, and an agent CLI with an initial login. Once installed, navigate to the TLDRack repository, run the necessary commands to configure the tool, and initiate your project. The interface presents a row for each agent session, featuring the generated name, a grade after each turn's completion, the terminal, and a live preview of the agent's workspace. The model's assessment determines whether additional work remains, significantly reducing the time spent reading. The tool supports three modes for executing substantial tasks: individual sessions, iterative agents with sub-agents, and LEAD with worker agents. Each method has its trade-offs in terms of cost and efficiency. While the model's evaluations are based solely on message content and not the actual work, grades serve as a starting point for review. The author shares a video tutorial and provides the repository link for those interested in exploring TLDRack further.",
  "summary": "In two weeks I sent my coding agents 1,523 prompts and read every closing message. I was fed up, so I put a small local model in front of them. It left me 696 to read. The tool is TLDRack. It is open source (MIT) and runs on macOS with Claude Code or Codex. Here is how to get it running and what you get. What it does TLDRack runs your coding agents side by side in tmux and shows every one of them…",
  "key_points": [
    "Author ran 1,523 coding prompts, reviewing all outcomes",
    "Integrated open-source tool TLDRack to assist agents",
    "TLDRack uses local model to grade agent replies"
  ],
  "editors_take": null,
  "illustration": null,
  "coverage": {
    "outlets": 1,
    "also_reported_by": []
  },
  "ai_generated": true,
  "disclaimer": "Summaries, key points and the editor’s take are written by software from other outlets’ reporting and may contain errors — always check the linked original."
}