Replies: 1 comment
|
We'd also love to see RunJam added to the official clients list on https://agentclientprotocol.com/get-started/clients. |
0 replies
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Uh oh!
There was an error while loading. Please reload this page.
Apologies upfront for the shameless self-promotion — I thought this community might be interested though.
I've been building RunJam, a free, open-source (MIT) desktop app that manages your coding agents from a single window. It's built on ACP top to bottom, which means you get to use Claude Code, Codex CLI and Gemini CLI side by side in one interface — with zero ACP rewrites, no per-agent config, and seamless session handoff between agents. It even auto-detects the agents already installed on your machine.
A few things I think this community will care about:
A session board instead of terminal chaos. Every running session becomes a card in one of four columns — idle, running, waiting for input, errored — so you can see what all your agents are doing at a glance. Drag to reorganize, click any card to jump straight back into that session, with full context. Sessions persist locally across restarts, so no more losing work when you close a window.
Cross-protocol model proxy. A built-in proxy translates between Anthropic, OpenAI and Gemini protocols in real time. Run GPT-4o inside Claude Code, Claude inside Codex, DeepSeek or Qwen inside Gemini CLI — model and tool are fully decoupled. A central model hub lets you configure each agent's model once, with aliases like fast / smart.
Local models, one click. Built-in llama.cpp management: download GGUF models (DeepSeek Coder, Qwen, Llama, Mistral) and run them locally, automatically wired through the same proxy — any agent, any model, zero API cost, fully offline.
Cost transparency. A per-session dashboard shows token usage, cost by model/agent/day, trend charts and cache hit rates. You actually see where your money goes.
Everything is local-first: data lives in ~/.runjam, API keys are stored in the system keychain, and nothing leaves your machine.
It's a bit of a passion project, and it's my personal daily driver for coding — would love to hear people's feedback.
Links: Website — https://runjam.app · Source & downloads — https://github.com/peintune/runjam · Platforms — macOS, Windows, Linux
All reactions