Why does it use hooks instead of simple session analysis?

Hooks provide 100% deterministic capture of every tool call and outcome, ensuring no patterns are missed, whereas standard session analysis is probabilistic.

How does the confidence scoring system work?

Instincts start with a provisional score (0.3) and increase as patterns are repeatedly observed or manually approved, reaching 'core behavior' status at 0.9.

What happens when an instinct 'evolves'?

The evolution process clusters related atomic instincts into higher-level assets like specialized background agents, reusable skills, or custom CLI commands.

Does this skill share my private source code?

No, all observations are stored locally on your machine. Only abstract 'instinct' patterns can be exported, and you have full control over what is shared.

Continuous Learning v2

Name: Continuous Learning v2
Author: saar120

bysaar120

0•

Productivity & Workflow

Transforms Claude Code session history into reusable instincts and automated skills using confidence-weighted observation hooks.

Continuous Learning v2 is an advanced learning architecture for Claude Code that captures user patterns, error resolutions, and stylistic preferences through deterministic execution hooks. By decomposing complex workflows into atomic 'instincts' with specific confidence scores, it allows Claude to progressively internalize your unique coding style and requirements. Over time, these patterns evolve into specialized commands, skills, or even background agents, creating a truly personalized AI assistant that grows more efficient with every interaction.

Key Features

01Automatic evolution of instinct clusters into full Claude Code skills, commands, and specialized agents

02Atomic 'Instinct' model with confidence scoring (0.3 to 0.9) to weight learned behaviors

030 GitHub stars

04Background pattern detection powered by Haiku to identify user corrections and repeated workflows

05Deterministic observation using PreToolUse and PostToolUse hooks for 100% reliable data capture

06Exportable and importable instincts for sharing learned behaviors across teams or devices

Use Cases

01Learning and automating repetitive multi-step troubleshooting workflows based on previous successful error resolutions

02Automatically capturing and enforcing specific project-wide coding style preferences such as functional vs. class-based patterns

03Building a library of personalized commands that evolve naturally from your most frequent tool usage patterns

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add saar120/skillgate-registry continuous-learning-v2

For use in Claude.ai and ChatGPT

Key Features

01Automatic evolution of instinct clusters into full Claude Code skills, commands, and specialized agents

02Atomic 'Instinct' model with confidence scoring (0.3 to 0.9) to weight learned behaviors

030 GitHub stars

04Background pattern detection powered by Haiku to identify user corrections and repeated workflows

05Deterministic observation using PreToolUse and PostToolUse hooks for 100% reliable data capture

06Exportable and importable instincts for sharing learned behaviors across teams or devices

Use Cases

01Learning and automating repetitive multi-step troubleshooting workflows based on previous successful error resolutions

02Automatically capturing and enforcing specific project-wide coding style preferences such as functional vs. class-based patterns

03Building a library of personalized commands that evolve naturally from your most frequent tool usage patterns