evolve

Analyze evidence and propose bounded AI prompt modifications for user approval.

2|Updated Apr 10, 2026
One-click install
npx skills add https://github.com/kevintelford/holdfast --skill evolve-kevintelford
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: evolve
Source: https://github.com/kevintelford/holdfast/tree/main/skills/evolve
Command: npx skills add https://github.com/kevintelford/holdfast --skill evolve-kevintelford

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve?

This Skill addresses the challenge of systematically proposing and applying bounded improvements to AI systems based on accumulated evidence, ensuring safe and controlled evolution.

Core Features & Use Cases

  • Evidence Analysis: Analyzes collected data and patterns to identify areas for improvement in AI outputs.
  • Proposal Generation: Crafts precise, bounded change suggestions for evolvable surfaces, specifying what changes, what remains fixed, and citing supporting evidence.
  • Application Workflow: Supports both pipeline and direct file editing modes for applying proposed evolutions, with a focus on user approval and versioning.

Quick Start

Use the evolve skill to analyze recent evidence and generate a proposal for improving the AI prompt or approach.

Frequently Asked Questions about evolve

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

FAQPage Schema
How do I propose bounded modifications to AI prompts using collected evidence?

Evidence-based AI evolution works by analyzing collected data patterns to identify output deficiencies, then crafting precise, bounded change proposals for evolvable surfaces. This mechanism ensures safe, controlled system improvements by requiring user approval before applying modifications.

What is the best way to apply incremental improvements to AI systems safely?

The best way to apply incremental improvements safely is through an application workflow that supports both pipeline and direct file editing modes. This approach focuses on user approval and versioning to control the evolution of AI prompts and approaches.

How do I generate an AI prompt improvement proposal from system output data?

You generate an AI prompt improvement proposal by analyzing recent evidence and output data to identify areas for improvement. The system crafts specific change suggestions that specify what changes, what remains fixed, and cite supporting evidence for the evolution.

Can I use direct file editing to apply changes to AI approaches without a pipeline?

Yes, you can use direct file editing to apply changes to AI approaches without a pipeline. The application workflow supports both pipeline and direct file editing modes for applying proposed evolutions, maintaining user approval and versioning controls.

What are the limitations of using bounded proposals for AI system optimization?

The limitation of using bounded proposals for AI system optimization is that modifications are constrained to specific, incremental changes rather than sweeping updates. This ensures safe evolution but requires multiple iterations to achieve large-scale system transformations.