Ralphy CLI
A CLI that loops Claude Code, Codex, OpenCode and other agents over a task list or PRD until everything is done
- 2,981 ★ on the repository
- Last commit
Broad access
Why this level
- Runs the coding agent with flags that skip action confirmation
- Can automatically create commits, merge branches and open PRs without manual review
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add cli/ralphyDetects 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.
- Runs the coding agent with flags that skip action confirmation
- Can automatically create commits, merge branches and open PRs without manual review
This is third-party code. Review the repository files before installing.
What it does
Ralphy takes a single task or a plan file (a markdown PRD, YAML, JSON, or a list of GitHub issues) and repeatedly runs the chosen coding agent until every item is marked done. It can run several agents in parallel, each in an isolated git worktree, with automatic branch merging or a PR per task. It supports eight engines to choose from, can skip tests and linting for speed, and can send Discord or Slack notifications when a session finishes.
Who it is for. For developers who want to run a long task list through a coding agent without manually restarting it after every step.
Good fit when
- You have a PRD or task list an agent can work through without supervision
- You need to run several independent tasks in parallel in separate worktrees
- You want an automatic PR for each completed task
- You want to quickly compare how different engines handle the same task
Not a fit when
- The task needs constant human oversight and intermediate decisions
- The repository is very large and git worktrees add too much overhead (the --sandbox flag helps but is not always enough)
Example request
Run Ralphy over PRD.md with three parallel Claude Code agents and create a PR for each finished taskLimitations
Requires the chosen engine's CLI to be installed and authenticated (Claude Code, Codex, and so on). It launches the agent with confirmation-skipping flags (such as --dangerously-skip-permissions), so the agent acts without step-by-step oversight. The bash version also needs jq, plus yq and bc for YAML tasks and cost calculation.
How to disable. Remove the global package with npm uninstall -g ralphy-cli, or delete the cloned folder containing ralphy.sh.
FAQ
Which engines are supported?
Claude Code, OpenCode, Cursor agent, Codex, Qwen-Code, Factory Droid, GitHub Copilot CLI and Gemini CLI, selected with a flag such as --opencode or --codex.
What happens in parallel mode?
Each agent gets its own git worktree and branch; without --create-pr, branches auto-merge back with AI-assisted conflict resolution, and with it, separate PRs are created instead.
Discussion
No questions or reviews yet.
Related
CLI toolsCLI that runs autonomous coding loops: a spec from GitHub, Linear or Notion, and an agent working until done.
The @itsaplan/runner CLI takes tasks from the It's a Plan tracker and hands them to Claude Code, Codex or opencode locally.
Orchestrator for Claude Code, Codex and Gemini CLI: YAML workflows that loop until your own quality gates pass
CLI dispatcher for agents: tasks run in the background and the agent sleeps on a question until you reply from a messenger.
A self-hosted control plane for running several agents: task dispatch, run review, spend tracking, connected via MCP, CLI or REST
groundcrew hands Linear and Jira tasks to local Claude Code, Codex and Cursor agents, each in its own worktree and sandbox
Comparison with similar tools
Cards that share the main search query, and how they differ.
| Tool | Official | Repository stars | Main difference |
|---|---|---|---|
| Ralphy (this card) | no | 2,981 | |
| Open Ralph Wiggum | no | 1,906 | the ralph command repeats one prompt for Claude Code |
| Ralph Loop by PageAI | no | 342 | a Ralph loop for long tasks |
README in short
The README describes Ralphy as an autonomous coding loop: the agent works through tasks from a PRD, YAML, JSON or GitHub issues until the list is done. It offers parallel execution with git worktrees or lightweight symlink-based sandboxes, per-task branches, auto-merge or PR creation, webhook notifications, and optional browser automation via agent-browser to verify UI. Both a TypeScript npm version and a bash script with identical features are available. MIT licensed.