YYLO CLI
An orchestrator CLI for coding agents: loops, typed tasks, receipt-backed merges and a skill set for Claude Code and Codex
- 66 ★ on the repository
- Last commit
- MIT
Broad access
Why this level
- The orchestrator launches an agent that runs commands and writes to the repository, with merges following receipts
- Token spend and cost grow with the number of loop iterations
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add cli/yyloDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewBroad accessThe tool can change data, run commands or handle secrets.
- The orchestrator launches an agent that runs commands and writes to the repository, with merges following receipts
- Token spend and cost grow with the number of loop iterations
This is third-party code. Review the repository files before installing.
What it does
YYLO runs a coding agent in a loop, manages tasks under typed rules and records receipts for every repository change. The `yylo` and `yy` commands are equivalent, and `ypl` abbreviates `yy pi --live`, opening an interactive agent session with streamed output and a status bar. After a run the orchestrator prints iterations, tool calls, cost and session ids. A first run needs no model provider: `yy init --task "..." --subagent pi` creates the `.juno_task/` workspace, and `yy doctor workspace` and `yy info --json` report its state. Models are picked with aliases such as `:sonnet`, `:opus` or `:gpt`, and the alias set depends on the subagent. Skills live in a separate `yylo-dev/yylo-skills` repository and install with `yy skills install`. There are seven: `artifact-yylo`, `ledger-tasks-yylo`, `plan-ledger-tasks-yylo`, `ralph-loop-yylo`, `understand-project-yylo`, `wiki-yylo` and `workflow-yylo`. They are copied into `.agents/skills`, `.claude/skills` and `.pi/skills`. The YYLO Ledger task store and the YYLO Benchmark evaluation ship as separate packages.
Who it is for. Developers who want a quick loop with a coding agent and project operators whose tasks, checks and merges must leave a trail.
Good fit when
- An agent needs several iterations on a task and you want to see cost and tool calls
- Project tasks are kept as typed records with checks and merge boundaries
- One repository is served by Claude Code, Codex and Pi at once and skills are needed in `.claude/skills` and `.agents/skills`
- Before a run the workspace must be checked with `yy doctor workspace` without changes
Not a fit when
- A one-off edit in a single file is needed: the orchestrator adds extra entities
- Node.js 20.10 or newer and npm are missing
- Only a task store or only a benchmark is needed: separate Ledger and Benchmark packages exist
Example request
Install YYLO, initialize the project with `yy init --task "Describe the onboarding path" --subagent pi` and check `yy doctor workspace`Limitations
It needs Node.js 20.10 or newer, npm and Git. The stable `@latest` channel lags the source: the README names a code version above the published one, and commands from newer versions may be missing in an installed build. Version 0.2.12 requires exactly Ledger 0.4.0, and other versions are rejected before delegation. Skills install over the network through `npx skills add`, falling back to a shallow clone on failure. The command refuses to replace hand-edited skill copies without `--force`. Model access depends on the chosen subagent and your keys or subscriptions.
How to disable. Remove the package with `npm uninstall --global @yylo/cli`, the project `.juno_task/` folder and the skill copies with the `-yylo` suffix from `.claude/skills`, `.agents/skills` and `.pi/skills`.
FAQ
How do I install YYLO?
Run `npm install --global '@yylo/cli@latest'` and check `yy --version`. Node.js 20.10 or newer, npm and Git are required. Prereleases are on the `@next` channel.
What is the difference between `yylo` and `yy`?
None: they are equivalent launchers. There is also `ypl`, shorthand for `yy pi --live`, which opens an interactive agent session.
How do I install YYLO skills for Claude Code?
Run `yy skills install`, optionally with `--version 2.1.1`. Seven skills are copied into `.claude/skills`, `.agents/skills` and `.pi/skills`. `yy skills status` shows their state.
Does the first run contact a model provider?
No. The demo run after `yy init` checks the installation, creates `.juno_task/` and emits a watch receipt with state `COMPLETED` without calling a model.
Is a separate Ledger required?
Not for ordinary work, but typed tasks and universal Record retrieval need YYLO Ledger at the exact required version. No substitute is installed automatically.
Discussion
No questions or reviews yet.
Related
CLI toolsSandbox CLI for agents: Claude Code and Codex run without prompts on a project copy, and changes apply after a diff.
A CLI and plugin: agent workflows live in the repository as Markdown and YAML with roles, scripts, pauses and branching.
One agent launches others in the background: Claude Code, Codex, Copilot CLI, Grok Build and Pi, with a model chosen per stage.
A CLI that runs parallel Claude Code sessions from your plan, each in its own git worktree, with stage checks and a knowledge base
A template sets up a multi-agent environment in a repository: Claude Code conducts Sonnet and Opus subagents and calls Codex CLI.
A terminal coding agent in one Rust binary that reads code, edits files and runs tests with any model of your choice.
Comparison with similar tools
Cards that share the main search query, and how they differ.
| Tool | Official | Repository stars | Main difference |
|---|---|---|---|
| YYLO (this card) | no | 66 | |
| Open Ralph Wiggum | no | 1,906 | a Ralph loop and nothing more, whereas YYLO adds typed tasks, merge receipts and a skill set |
| Ralph | no | 22,226 | an implementation of the agent loop idea driven by a PRD, whereas this is an orchestrator with tasks, merge checks and several subagents |
README in short
The README covers quick start, version channels, skill installation, model aliases, a beginner workflow, typed tasks and merging, managed workflows and maintainer roles.