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.
npx agentmods add instructions/kylemclaren/ralph/claude-mdgit clone --depth 1 https://github.com/kylemclaren/ralphWrote 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.
[](https://agentmods.dev/instructions/kylemclaren/ralph/claude-md)<a href="https://agentmods.dev/instructions/kylemclaren/ralph/claude-md"><img src="https://agentmods.dev/badge/instructions/kylemclaren/ralph/claude-md.svg" alt="Measured on agentmods" height="20"></a>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.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.00656 | $0.00656 |
| Opus 5 | $0.00328 | $0.00328 |
| Sonnet 5 | $0.00131 | $0.00131 |
| Haiku 4.5 | $0.00066 | $0.00066 |
Grade A, and why
ralph CLAUDE.md 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 5d 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.
How it starts
The opening of the file, as written. The whole thing — 61 lines — stays where its author put it; the contents beside it link to each section on GitHub.
CLAUDE.md
This file provides guidance to Claude Code (claude.ai/code) when working with code in this repository.
Build & Development Commands
mise run build # Build ralph binary
mise run test # Run all tests
mise run lint # Run golangci-lint
mise run install # Install to GOPATH/bin
# Run a single test
go test -v -run TestName ./internal/prd/
# Run tests for a specific package
go test -v ./internal/agent/
Architecture
Ralph is an autonomous AI coding loop that repeatedly runs AI coding agents (Claude Code, Amp, OpenCode, Codex) until all tasks in a PRD are complete. Each iteration is a fresh context window with memory persisting via git and text files.
Core Loop (internal/loop/loop.go)
The main execution loop that:
- Loads PRD, progress log, and prompt template
- Finds highest priority pending story
- Renders the prompt with current state
- Executes the configured AI agent
- Checks for completion marker (
<promise>COMPLETE</promise>) - Repeats until all stories pass or max iterations reached
Key Packages
-
internal/agent- Agent abstraction for executing AI tools. Handles command building, timeout, environment variables, and output capture. Agents receive prompts via-pflag. -
internal/prd- PRD (Product Requirements Document) management. UserStories have ID, title, acceptance criteria, priority, and passes boolean. NextStory() returns highest priority pending story. -
internal/config- Configuration via Viper. Supports ralph.yaml file, RALPH_ env vars, and CLI flags. Config precedence: flags > env > file. -
internal/hooks- Lifecycle hooks (onStart, onIteration, onComplete, onFailure) that run shell commands at key points. -
internal/prompt- Go text/template rendering for agent prompts. TemplateData includes PRD JSON, progress content, and story counts. -
internal/claudecode- Environment variable bridge. Sets RALPH_* env vars so Claude Code hooks can access Ralph state (iteration, story ID, etc.).
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.
- 5d ago First seen · 61 lines · 656 tokens per session scan A 1c4bb4a53853
ralph CLAUDE.md is an instructions file published in the GitHub repository kylemclaren/ralph (10 stars, last pushed 7mo ago), licensed MIT. It adds 656 tokens to every session, about $0.0033 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.
Other instructions, from other repositories
ralphex CLAUDE.md
Claude Code instructions for umputun/ralphex, covering ralphex, llm documentation, build commands, updating dependencies and project structure.
ralphy CLAUDE.md
Claude Code instructions for michaelshimeles/ralphy, covering claude code guidelines for ralphy, code change philosophy, keep changes small and focused, task prioritization and code quality standards.
ralph-desktop AGENTS.md
AGENTS.md instructions for liuxiaopai-ai/ralph-desktop, covering agents.md, project overview, repo layout, dev commands and conventions.
ralphctl CLAUDE.md
Claude Code instructions for wcygan/ralphctl, covering claude.md, project overview, build & test, cli commands and dependencies.
rooda AGENTS.md
AGENTS.md instructions for jomadu/rooda, covering agent instructions, issue tracking, quick reference, landing the plane (session completion) and work tracking system.
ralph-planner CLAUDE.md
Claude Code instructions for vavasilva/ralph-planner, covering ralph-planner, location, structure, commands and key features.