Skills
.Work. You
Rest

exo:ralph

Official

Enable iterative refinement loops with Ralph

Software Engineering#swarm#agent-loop#scorers#ralph#iterative-refinement#reflectors#streaming-events
Authormidsphere-ai
Version1.0.0
Installs0

System Documentation

What problem does it solve?

Builds and orchestrates iterative refinement loops for autonomous agents using Ralph — RalphRunner, enabling a defined 5-phase cycle (Run/Analyze/Learn/Plan/Halt) with support for scorers, reflectors, stop conditions, and streaming events, plus integration points for Swarm via RalphNode.

Core Features & Use Cases

  • Orchestrates a 5-phase loop (Run, Analyze, Learn, Plan, Halt) using RalphRunner to manage execution, scoring, and feedback.
  • Provides from_agent() factory to wire an Agent's run and run.stream() into a Ralph loop for seamless integration.
  • Supports Swarm integration via RalphNode to embed refined loops into multi-agent workflows; streams RalphIterationEvent and RalphStopEvent for observability.
  • Configurable stop conditions, scoring and reflection, enabling quality-gated iterative improvements across iterations.

Quick Start

Provide an Agent, create a RalphRunner via from_agent with scorers and a stop condition, then call run to commence iterative refinement.

Dependency Matrix

Required Modules

None required

Components

Standard package

💻 Claude Code Installation

Recommended: Let Claude install automatically. Simply copy and paste the text below to Claude Code.

Please help me install this Skill:
Name: exo:ralph
Download link: https://github.com/midsphere-ai/exo/archive/main.zip#exo-ralph

Please download this .zip file, extract it, and install it in the .claude/skills/ directory.
View Source Repository

Agent Skills Search Helper

Install a tiny helper to your Agent, search and equip skill from 471,000+ vetted skills library on demand.