Skip to content
Agentmart
Submit a tool

ralph-loop

Start a Ralph Loop — autonomous work-until-done via Stop-hook continuations.

Skill by mihazs

IndexedMIT★ 11
Write a review

Install

Works with Claude Code, Claude apps, Codex, Cursor, Gemini CLI, GitHub Copilot, VS Code, OpenCode, Goose, Windsurf. Read the repository before you give any tool access to your data.

npx agentmarket install skill/mihazs-ralph-loop

Asks which of your tools to add it to (Claude Code, Codex, Cursor, VS Code, Windsurf, Gemini CLI, Claude Desktop) and shows the change first. See https://agentmart.dev/cli.

Get the weekly diff

The weekly diff: one email a week with new agent tools, notable updates and grade changes, plus security alerts for tools you follow. Double opt-in, one-click unsubscribe.

We use your email only for this newsletter. Privacy

About

Start a Ralph Loop — autonomous work-until-done via Stop-hook continuations. Use when the user says ralph loop, /ralph-loop, or wants the agent to keep going until it emits a completion promise tag. Pair with cancel-ralph to stop.

Use cases: agent-building

No reviews yet. Installed it? Reviews come from verified installers only. How reviews work

Permissions

What this skill can do and what it asks for, from its SKILL.md. Checked Oct 25, 2026.

Tools
No tools found
Secrets
None asked for
Install scripts
None found

Network hosts it talks to 1

  • codeload.github.com · found in install command

Files it touches

  • ~/.claude/skills/ralph-loop

What's inside

Skill name
ralph-loop
Description
Start a Ralph Loop — autonomous work-until-done via Stop-hook continuations. Use when the user says ralph loop, /ralph-loop, or wants the agent to keep going until it emits a completion promise tag. Pair with cancel-ralph to stop.
License
MIT
Instructions
49 lines in SKILL.md · view on GitHub

Files (1)

  • SKILL.md

SKILL.md (first lines)

# Ralph Loop

## Start

User runs:

```text
/ralph-loop "your task description" [--completion-promise=DONE] [--max-iterations=100] [--strategy=continue|reset]
```

Or invoke this skill and state the task in the same message.

## Behavior (Grok hooks)

1. `UserPromptSubmit` writes `.omg/ralph-loop.local.md` in the workspace and injects loop instructions.
2. You work until the task is **fully** complete.
3. When done, output: `<promise>DONE</promise>` (or your custom `--completion-promise` text).
4. If you stop without that tag, the **Stop** hook blocks exit and injects a continuation prompt (up to `max-iterations`, default 100).
5. Pending todos can still block stop when no Ralph loop is active (`stop-verify-pending`).

## Rules

- Do not emit the completion promise early.
- Use todos for multi-step work.
- Each iteration must make real progress.
- Do not ask the user to continue or start the "next phase" — the Stop hook auto-continues until the promise tag or `/cancel-ralph`.

## Ultrawork variant

For verified completion (Oracle subagent required), use **`ulw-loop`** instead:

```text
/ulw-loop "same task" [--max-iterations=500]
```

See the oh-my-grok `ulw-loop` skill (`grok inspect` for path).

## Cancel

```text
/cancel-ralph
```

Or use the `cancel-ralph` skill. Also cancels an active ultrawork loop.

## State file

Workspace-relative: `.omg/ralph-loop.local.md` (oh-my-grok; omo uses `.omo/`).

From mihazs/oh-my-grok, MIT.

More from mihazs/oh-my-grok

Signals

Stars
11
Forks
5
Open issues
0
Last commit (this folder)
Jun 3, 2026 (4 months ago)
Last release
Jun 3, 2026 (4 months ago)

From GitHub (fetched Oct 24, 2026).

Details

License
MIT
First seen
Oct 24, 2026
Last checked
Oct 24, 2026

More from mihazs

Cite this page

ralph-loop skill. Agent Mart. https://agentmart.dev/skills/mihazs-ralph-loop (last updated 2026-10-24)