continuous-learning-v2

Analyze Claude Code sessions to convert recurring patterns into atomic instincts.

Updated Jan 30, 2026
One-click install
npx skills add https://github.com/CobaltSato/avalanche-build-games-tool-kit --skill continuous-learning-v2-cobaltsato
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: continuous-learning-v2
Source: https://github.com/CobaltSato/avalanche-build-games-tool-kit/tree/main/.claude/skills/continuous-learning-v2
Command: npx skills add https://github.com/CobaltSato/avalanche-build-games-tool-kit --skill continuous-learning-v2-cobaltsato

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) components.

What problem does it solve?

Instinct-based learning system that observes Claude Code sessions via hooks, creating atomic instincts with confidence scoring and evolving them into skills, commands, and agents to accelerate pattern discovery and automation.

Core Features & Use Cases

  • Atomic instincts with confidence weights enable incremental learning and safe evolution into more complex capabilities.
  • Observations captured by PreToolUse and PostToolUse hooks provide reliable data for pattern detection.
  • Evolving instincts into clusters, skills, commands, and agents supports scalable automation and expert guidance across coding workflows.
  • Use case: A developer captures a recurring workflow and the system evolves it into a reusable skill that can be invoked as a command or integrated into automation.

Quick Start

Enable observation hooks, initialize the homunculus directories, and start the observer to begin capturing sessions for instinct generation.

Frequently Asked Questions about continuous-learning-v2

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

FAQPage Schema
How do I capture recurring patterns from Claude Code sessions into reusable automation?

You can capture recurring patterns from Claude Code sessions by enabling PreToolUse and PostToolUse hooks. These hooks observe sessions, generate atomic instincts with confidence scoring, and evolve them into deployable skills, commands, or agents.

What are atomic instincts in continuous learning and how do they work?

Atomic instincts are incremental learning units created from observed session patterns. Each instinct carries a confidence weight that determines its reliability, allowing safe evolution into more complex automation capabilities like skills or commands.

How do I set up session observation hooks to start generating instincts?

To set up session observation hooks, initialize the homunculus directories under the ~/.claude/homunculus path and start the observer. This enables PreToolUse and PostToolUse hooks to capture sessions for instinct generation.

Can I evolve captured workflow patterns into deployable skills and commands?

Yes, you can evolve captured workflow patterns into deployable skills and commands. The system clusters atomic instincts based on confidence scores and transforms them into reusable skills, commands, or agents for scalable automation.

Does this instinct-based learning system require any external dependencies?

No external dependencies are required. The system operates locally, storing observations and evolved instincts under the ~/.claude/homunculus path, relying solely on Claude Code hooks for data capture.

What is the best way to automate pattern discovery across developer workflows?

The best way to automate pattern discovery is using an instinct-based learning system that observes sessions via hooks, detects recurring patterns, and evolves them into atomic instincts with confidence scoring for safe deployment as automation.