ralph

ralph is a skill for Claude Code from mischasigtermans/claude-ralph. It costs 79 tokens per session (2,686 once invoked), scanned A, original, MIT.

A planning skill for the autonomous Ralph coding loop, which repeatedly works through a project from a written plan. It interviews the user, writes a project brief and story list, and sets the planner's review schedule.

In plain words
What is it for?
Use it to define project scope, record requirements in `.ralph/brief.md`, create work stories, choose how often progress is reviewed, and migrate older `.ralph/` folders.
Why use it?
It turns a broad coding request into a shared plan that each loop iteration can follow. It also prevents the setup session from starting implementation prematurely.

Skill for Claude Code

Written for Claude Code: shipped in a Claude Code plugin. Also seen: mentions CLAUDE.md.

Part of the ralph plugin — 1 skill shipped together

Good fit Use it to define project scope, record requirements in .ralph/brief.md, create work stories, choose how often progress is reviewed, and migrate older .ralph/ folders.

Compare 6 skills from other repositories ↓
Install with agentmods
npx agentmods add skills/mischasigtermans/claude-ralph/ralph
Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

Any agent
npx skills add mischasigtermans/claude-ralph --skill ralph
Clone the repo
git clone --depth 1 https://github.com/mischasigtermans/claude-ralph

Made for: Claude Code.

Or install ralph, the plugin that ships this one along with the rest of its 1 skill.

Wrote this? Show the measurements

A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.

agentmods badge for ralph

README.md
[![agentmods](https://agentmods.dev/badge/skills/mischasigtermans/claude-ralph/ralph/github.svg)](https://agentmods.dev/skills/mischasigtermans/claude-ralph/ralph)
Your own site
<a href="https://agentmods.dev/skills/mischasigtermans/claude-ralph/ralph"><img src="https://agentmods.dev/badge/skills/mischasigtermans/claude-ralph/ralph/github.svg" alt="Measured on agentmods" height="20"></a>

Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.

agentmods 80×15 button for ralph

Your own site · 80×15
<a href="https://agentmods.dev/skills/mischasigtermans/claude-ralph/ralph"><img src="https://agentmods.dev/badge/skills/mischasigtermans/claude-ralph/ralph.svg" alt="Reviewed on agentmods" width="80" height="20"></a>
Per session 79 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 2,686 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. A grade says what 26 rules found in the file — not that it is safe.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5.1 $0.00079 $0.02686
Opus 5 $0.00039 $0.01343
Sonnet 5 $0.00016 $0.00537
Haiku 4.5 $0.00008 $0.00269

Measured 8d ago against content hash 5837ff157dc6, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-09, from the pricing page.

Security

Grade A, and why

ralph scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 8d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

skills/ralph/SKILL.md · 262 lines

How it starts

The opening of the file, as written. The whole thing — 262 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Ralph Skill (v2)

Your job in this skill is to scope Ralph's autonomous coding loop. Write a prose brief that survives across every loop iteration, draft the story list, and decide how often the loop's planner agent should review progress. Then exit. The user runs ralph separately.

CRITICAL: NEVER start implementing within this session. After setup is complete, tell the user to run ralph from a terminal.

The brief is the contract

The single most important thing you produce is .ralph/brief.md. It is prose, not bullets. Each loop iteration loads it as the plan-time context. The builder agent reads it before picking a story; the planner agent treats it as read-only ground truth.

Steve Jobs's exit criterion: if you cannot describe the work in two paragraphs, the planning isn't done yet. If you find yourself writing a four-page brief, you don't understand the scope well enough. Go back and ask one more question, or push back to narrow the work.

Step 1: Discover context (silent, no questions yet)

Before asking the user anything, do this on your own:

  1. Read CLAUDE.md if it exists in the project root. Note any conventions, patterns, or external context sources mentioned (issue trackers, spec dirs, dashboards, design docs).
  2. Read any context files referenced by CLAUDE.md that look directly relevant to the user's request.
  3. If a referenced source is reachable via an MCP server loaded in this session (Linear, GitHub, Notion, Sentry, etc.), and the user's job description references something matching it (e.g. "PROJ-123" when CLAUDE.md mentions Linear team PROJ), fetch the relevant items quietly. If no MCP for the source is loaded, skip. Don't ask the user to set one up.
  4. Briefly explore the codebase structure (top-level dirs, package manager, framework signals) so you can write a brief that uses the project's own vocabulary.

If .ralph/ already exists, check what's there before proceeding (see "Existing .ralph/" below).

Read the full file on GitHub · 262 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 8d ago First seen · 262 lines · 79 tokens per session scan A 5837ff157dc6

Subscribe to this mod's changes

ralph is a skill published in the GitHub repository mischasigtermans/claude-ralph (1 stars, last pushed 3mo ago), licensed MIT. It adds 79 tokens to every session and 2,686 once invoked, about $0.0004 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.

Related

Other skills, from other repositories