capability-evolver

Name: capability-evolver
Availability: InStock
Author: liutao96

Analyzes runtime history to generate auditable gene/capsule evolution prompts and artifacts via GEP Protocol.

Updated Feb 14, 2026

Search Tags:#automation #security #ai #evolution #auditable #gep

Ready to run

npx skills add https://github.com/liutao96/skill --skill capability-evolver-liutao96

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: capability-evolver
Source: https://github.com/liutao96/skill/tree/main/capability-evolver
Command: npx skills add https://github.com/liutao96/skill --skill capability-evolver-liutao96

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires dotenv.

What problem does it solve?

Evolver addresses the challenge of turning ad hoc prompt tweaks into auditable, reusable evolution assets for AI agents. It analyzes runtime history to identify failures and opportunities for improvement, and then drives protocol-constrained evolution to produce actionable prompts and artifacts.

Core Features & Use Cases

Auto-Log Analysis: scans logs and memory for signals to guide evolution.
Self-Repair Guidance: emits repair-oriented directives from signals to stabilize operation.
GEP Protocol: provides a standardized, auditable evolution workflow with gene/capsule assets.
Skill Store & Looping: supports offline local execution and optional hub-connected sharing of skills.
Safety & Governance: enforces safety checks, sandboxed validation for patches, and explicit review steps before applying changes.

Quick Start

Run node index.js to start an autonomous evolution cycle using the local memory and gep assets.

Frequently Asked Questions about capability-evolver

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

FAQPage Schema

How do I make ad hoc prompt tweaks auditable for AI agents?▼

You can make prompt tweaks auditable for AI agents by analyzing runtime history to identify failures and driving protocol-constrained evolution to produce actionable, reusable artifacts. This replaces manual adjustments with structured assets.

How does the GEP protocol standardize AI agent self-evolution?▼

The GEP protocol standardizes AI agent self-evolution by providing a structured, auditable workflow that utilizes gene and capsule assets. This enforces safety checks and sandboxed validation before applying any prompt or code changes.

Do I need a Git repository to automate self-repair guidance for my AI agent?▼

Yes, automating self-repair guidance requires a Git repository with the Evolver code, a memory store at memory/ and assets/gep/*, and Node.js version 18 or higher to run the evolution cycle.

Can I share evolved skills offline without connecting to a hub?▼

Yes, you can execute offline local runs using the local memory and gep assets without a hub connection. Optional hub-connected sharing is supported via the A2A_HUB_URL environment variable when networked sharing is needed.

What safety policies are required when applying patches to AI agent prompts?▼

Required safety policies for patching AI agent prompts include restricting shell commands, mandating explicit review steps, and enforcing sandboxed validation before applying changes to ensure deterministic self-improvement and governance.