ralph

Iteratively complete development tasks with automated verification and architect sign-off.

Updated Apr 10, 2026
One-click install
npx skills add https://github.com/lpjhelder/kronon-forge --skill ralph-lpjhelder
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: ralph
Source: https://github.com/lpjhelder/kronon-forge/tree/main/skills/ralph
Command: npx skills add https://github.com/lpjhelder/kronon-forge --skill ralph-lpjhelder

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Ralph prevents incomplete or superficially "done" work by running a persistent iterative loop that forces implementation to continue until fresh verification evidence (tests, builds, lint) and architect sign-off confirm true completion. It avoids silent failures, skipped tests, and scope reduction that commonly occur in multi-step development tasks.

Core Features & Use Cases

  • Persistent state & snapshots: Save and resume grounded context for long-running tasks across iterations.
  • Parallel delegation: Fire specialist agents concurrently at appropriate tiers (LOW, STANDARD, THOROUGH) for independent workstreams.
  • Mandatory verification & cleanup: Run automated verification (tests/builds/lint), require architect approval, run ai-slop-cleaner on changed files, and re-run regressions before completion.
  • Use Case: Finish a multi-file refactor that must pass full test suites and receive architect approval before merging.

Quick Start

Invoke the ralph skill with a clear task statement, for example: ralph finish the authentication feature and verify all tests and builds pass.

Frequently Asked Questions about ralph

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I ensure complex multi-iteration code changes finish and pass verification?

To ensure complex code changes finish, use a persistent iterative loop that forces implementation to continue until fresh verification evidence like tests, builds, and lint confirm true completion. This prevents silent failures and skipped tests in multi-step development tasks.

What is the best way to manage state persistence for long-running fixes?

State persistence for long-running fixes is managed by saving and resuming grounded context snapshots across iterations. This guarantees tasks requiring multiple iterations retain their progress and continue until automated test regressions and architect sign-off are achieved.

Can I run parallel agent delegation for independent workstreams?

Yes, you can fire specialist agents concurrently at appropriate tiers like LOW, STANDARD, or THOROUGH for parallel delegation of independent workstreams. This ensures complex multi-file refactors receive the required automated verification and architect approval before merging.

Does automated verification require architect approval before cleanup?

Automated verification requires running tests, builds, and lint, followed by mandatory architect approval. After sign-off, an ai-slop-cleaner runs on changed files and re-runs regressions to guarantee no superficially done work passes completion.

Why does my multi-file refactor skip tests and reduce scope?

Multi-file refactors skip tests and reduce scope due to silent failures from incomplete task execution. An iterative persistence loop prevents this by forcing implementation to continue until fresh verification evidence and architect sign-off confirm true completion.