bond

English

Español

English

Español

Appearance

bondYour coding agents, working together

pi, Claude Code, Codex, Copilot, opencode, Crush or Gemini. Use whichever you like, configure them once per repo and open them in Herdr. Licence to code.

Start here
View on GitHub
Download

bond setup

You do it once on your machine. It helps you with Herdr, brings in your models (Ollama, LM Studio, OpenAI or your own gateway) and gets the agents you use ready.

bond init

Once per repo. Pick your agents and tools, and bond writes each one's config from a single .bond/bond.toml. Anything you edited by hand stays untouched.

bond

The everyday part. Opens your repo in Herdr with your favourite agent. Lots to do? bond team splits branches across several agents, each in its own worktree.

Team catalogs

Your team's skills, commands and MCP servers in their own repo, with the version pinned. Everyone works with the same set, whichever agent they use.

What is this about? ​

If you use coding agents, you have probably been here: Claude Code wants a CLAUDE.md, pi reads AGENTS.md, Copilot looks for MCP servers in one place and opencode in another. And the moment you want two agents working at once, each on its own branch, you end up with five terminals and no idea which one is waiting for you to answer.

bond sorts that out. You declare once what you want in your repo (agents, MCP servers, skills, commands) and it translates it into each agent's format. Then it opens your workstation in Herdr, a terminal multiplexer built for agents: panes stay alive even if you close the terminal, and the sidebar shows at a glance who is working, who is done and who needs you.

Install it ​

sh
brew install jmanuelcorral/tap/bond                                                    # macOS and Linux
curl -fsSL https://raw.githubusercontent.com/jmanuelcorral/bond/main/install.sh | sh   # without Homebrew

On Windows: powershell -ExecutionPolicy Bypass -c "irm https://raw.githubusercontent.com/jmanuelcorral/bond/main/install.ps1 | iex".

Then run bond setup and off you go. We walk you through it in Getting started.

Messages in Spanish

bond's own messages (command output, bond doctor rows, status names such as trabajando or lista) are in Spanish. These docs gloss them in English where they appear.

See it in action ​

Start with this one. You open bond with Claude alone in its pane and ask it, in a single sentence, for a very simple website: make a plan, split it between two pi agents with bond team and act as the orchestrator. Everything else happens on its own, inside bond: the minions each write their page, the live table shows how they are doing, Claude reviews and sends two fixes back, merges into main and tidies up the team. No cuts: it takes as long as it takes.

Claude orchestrates two pi agents: plan, split, four panes at once, review with fixes, merge and cleanup.

And four more, short ones, on fluzo, a real Rust project:

Setting up a repo with bond init
The workstation in Herdr
A team of agents with bond team
Team catalogs

The code, on GitHub ​

bond is free software (MIT). The source, the releases and the changelog live at github.com/jmanuelcorral/bond. If something does not work or you have an idea, open an issue: it is much appreciated.