Coding

ralph

Try it

Guide and conventions for running the ralph autonomous loop tool (frankbria/ralph-claude-code) with Claude Code. Covers wrapper setup that avoids Circuit Breaker false positives, PUBLIC repo language guards for autonomous posting, and the workspace `.ralph/` artifact layout.

What it does

Reusable knowledge for running (an autonomous loop that drives to completion) against the Claude Code CLI. This skill is a **knowledge-only** reference — it ships no hooks and no agents. If you also want always-on enforcement (SessionStart guard injection, PreToolUse misuse blocking), you'll need…

The skill document

ralph

Reusable knowledge for running frankbria/ralph-claude-code (an autonomous loop that drives claude to completion) against the Claude Code CLI. This skill is a knowledge-only reference — it ships no hooks and no agents. If you also want always-on enforcement (SessionStart guard injection, PreToolUse misuse blocking), you'll need to build that yourself for your own workspace; this skill only documents the patterns.

When to use

  • Setting up a new workspace for Ralph (.ralphrc, claude-wrapper-ralph.sh)
  • Debugging Ralph Circuit Breaker (CB) OPEN states
  • Authoring autonomous-loop prompts that may post to PUBLIC GitHub repos
  • Understanding the .ralph/ artifact layout (fix_plan.md, docs/generated/)

When not to use

  • For OMC's PRD-driven loop (a separate tool also named "ralph"), use that tool's own guide
  • For one-shot tasks without persistence — Ralph adds overhead

Ralph's circuit breaker counts permission denials. Claude Code's permission system splits commands at |, &&, etc. without honoring shell quotes, so commands like gh ... --jq '.[] | select(...)' get denied even though they are safe. A wrapper that forces --permission-mode bypassPermissions works around this while your own destructive-command guard hooks (if any) stay active.

# ~/.ralphrc (per workspace)
CLAUDE_CODE_CMD="claude-wrapper-ralph.sh"

The wrapper script itself (e.g. at ~/.local/bin/claude-wrapper-ralph.sh) and any PreToolUse destructive-command guard are outside this skill's scope — they are workspace/operator-specific enforcement, not generic knowledge.

PUBLIC repo language guard (autonomous loops)

Ralph autonomous loops MUST translate non-English content to English before posting to any GitHub PUBLIC repo (issues, PRs, comments). A --no-confirm flag does not exempt this rule. Per-call procedure:

  1. gh repo view --json isPrivate -q '.isPrivate' — false means PUBLIC
  2. Scan the body for non-ASCII script characters; if any are found, translate to English first
  3. Localized issue-drafts are pre-publish working files only — never --body-file them directly to a PUBLIC repo

Artifact paths

PathPurpose
/.ralphrcper-workspace Ralph config (CLAUDE_CODE_CMD, retries, ALLOWED_TOOLS)
/.ralph/fix_plan.mdrunning task checklist (autonomous loop and supervisor share this file)
/.agents/docs/generated/plan-*.mdper-task plan artifacts
/.agents/docs/generated/research-*.mdper-task research artifacts

"Workspace" here means the directory that contains .ralph/ — typically a parent of one or more git repos, not the repo itself. Search for .ralph/ at workspace level, not project level.

Installation

clawhub install ralph

or add this marketplace and install the bundling plugin (this skill ships inside the labs plugin alongside other auxiliary skills):

claude plugin marketplace add https://github.com/es6kr/skills
claude plugin install labs@es6kr-skills

Either path provides this SKILL.md for reference only. It intentionally has no SessionStart guards and no PreToolUse hooks — those are enforcement concerns you add on top for your own workspace.

Related skills

Stores durable facts in a categorized, plain-markdown vault on disk, alongside your agent's built-in memory.

by Iván559 installs18 stars

Generate and edit Draw.io, Mermaid, and Excalidraw diagrams from natural language using a structured JSON spec.

by nssa.io1.0k installs47 stars

Join a video meeting as an AI bot with voice, avatar, and screenshare across four operating modes.

by johnpatternai22 installs8 stars

Find why your productivity system keeps failing, then apply the smallest fix — capacity math, bottleneck routing, durable local notes.

by Iván855 installs69 stars

Fetch raw ad creative, app, ranking, and revenue data from AdMapix as structured JSON.

by fly0pants4.3k installs296 stars

Run Git operations — commits, branches, merges, rebases, conflict resolution, and recovery — with safety rules enforced.

by Iván535 installs31 stars

More from drumrobot

Browse all skills

Run a 4-stage research → plan → review → TDD-implement workflow for code changes with optional PR capture.

by drumrobot27 installs

Manage Claude Code sessions: lookup, search, repair, migration, compression, and cleanup in one command surface.

by drumrobot26 installs

Manage Git worktrees, ghq clones, SourceGit registration, and multi-account credentials with safety checks.

by drumrobot26 installs

Diagnose agent mistakes, harden the prompt that caused them, and finish the original task.

by drumrobot25 installs

Convert plans and code changes into GitHub issues and PRs, with auth handling, dependency tracking, and pre-publish scans.

by drumrobot18 installs

Convert functionality into Claude Code automations (Agent, Skill, Rule, Hook, Command)

by drumrobot17 installs