leonidb

Jean — Development skill for Claude Code

Development community

Runs a group of Claude Code agents on one project, with one of them in charge.

How to install Jean

This entry records only its repository, not the path inside it, so there is no exact command to give. Open leonidb/jean and copy the folder into ~/.claude/skills/, or the file into ~/.claude/agents/.

What Jean does

Runs a group of Claude Code agents on one project, with one of them in charge. Local, event-sourced, one person, a handful of agents.

Alternatives in Development

  • Om Prep 1on1 — Prep for an upcoming 1:1 — load person context, surface open items, suggest agenda based on vault state 4.6k ★
  • Purpose — Orient yourself in pi-vs-cc — a collection of Pi coding agent extensions and agent specs that progressively de 527 ★
  • Custom Dimensions — Custom Dimensions and Event Parameters 242 ★

README

Jean

[![ci](https://github.com/leonidb/jean/actions/workflows/ci.yml/badge.svg)](https://github.com/leonidb/jean/actions/workflows/ci.yml) [![license](https://img.shields.io/badge/license-Apache--2.0-blue)](LICENSE)

Jean /ʒɑ̃/ (zhon) runs a group of Claude Code agents on one project, with one of them in charge.

A **dojo** is where a group trains under one teacher; here it is a project, the agents working on it, and the record of everything that passed between them. The **sensei** is the agent in charge: you talk to it, and it turns what you ask for into tasks and hands them out. A **worker** is an agent that takes a task, does it, and reports back.

You tell the sensei what you want. It writes a task, picks a worker, and sends it over. The worker does the work and reports back to the sensei, which has the result when you ask — and if the worker goes quiet, the sensei hears about that too. Every task, message and report passes through the dojo's log as it happens, so when you come back you read what happened, in order. What the agents learn while they work — about the project, and about how you work — accumulates in the dojo's **library**: a wiki built from what they memorize, which every agent reads, so the dojo grows more tuned to your work the longer it runs.

Everything runs on your machine: the agents, the log, and the server they share. One person, one project, a handful of agents. Claude Code is the reference runtime, not a limit of the design. The dojo — its log, board, mailbox and the protocol agents speak — is tied to no provider; Codex and OpenCode are planned next.

New to Jean? Start with [your first session](docs/running.md#your-first-session).

Jean's capabilities as four stacked layers — agents and their playbooks on top, then accountability, then knowledge, all resting on an append-only event log, with a person beside the stack connected </div> </section>  </div> <section class=

Related Skills