self-evolution-module

Orchestrate reflection-driven retry loops that evolve prompts, tools, and workflows.

43|2|Updated Mar 26, 2026
One-click install
npx skills add https://github.com/curated-skills/LinguaClaw --skill self-evolution-module
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: self-evolution-module
Source: https://github.com/curated-skills/LinguaClaw/tree/main/harnesses/modules/self-evolution
Command: npx skills add https://github.com/curated-skills/LinguaClaw --skill self-evolution-module

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This Skill solves repeated agent failures by forcing a structured reflection-driven retry loop that evolves the prompt, tools, and workflow based on observed outcomes instead of cosmetic changes.

Core Features & Use Cases

  • Prompt evolution to sharpen success conditions, refine framing, and better decompose tasks when understanding or instructions are the bottleneck.
  • Tool evolution to improve observability, add specialized helpers, or wrap/replace tool interfaces when execution is brittle or inefficient.
  • Workflow evolution to change the process graph (gates, verifiers, sequencing, or parallel branches) when sequencing or missing checks cause instability.
  • Use case: An agent repeatedly produces incomplete solutions because required acceptance criteria are not actually verified; this Skill drives attempt-by-attempt redesign until the acceptance gate is satisfied.

Quick Start

Ask the agent to run a reflection-driven retry loop for the task, using the attempt cap (default 5) and only stopping when the task’s explicit acceptance gate is met.

Frequently Asked Questions about self-evolution-module

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

FAQPage Schema
How do I stop an agent from making cosmetic changes on repeated failures?

To stop cosmetic changes on repeated agent failures, use a reflection-driven retry loop that forces material evolution of the prompt, tools, or workflow based on observed trajectories rather than minor edits.

What is a reflection-driven retry loop for agent tasks?

A reflection-driven retry loop is an orchestration pattern that improves subsequent agent attempts by reflecting on observed failures, evolving prompt framing, tool surfaces, or workflow sequencing until an explicit acceptance gate is satisfied.

How do I evolve agent workflows when sequencing causes instability?

To evolve agent workflows when sequencing causes instability, apply workflow evolution to change the process graph, adding gates, verifiers, or parallel branches until the acceptance gate is satisfied or the attempt cap is exhausted.

Can I use prompt evolution to sharpen agent success conditions?

Yes, you can use prompt evolution to sharpen agent success conditions, refine framing, and decompose tasks when understanding or instructions are the bottleneck, ensuring material changes by the second attempt.

Why does my agent stop retrying before the acceptance gate is met?

An agent stops retrying before the acceptance gate is met when it hits the configured attempt cap, which defaults to 5, or if it lacks a structured reflection loop to drive continued retries until success.