Why use hooks instead of standard triggers for learning?

Hooks are 100% reliable and deterministic, meaning every tool usage is analyzed for potential learning opportunities, whereas standard triggers can be inconsistent.

How does version 2.1 handle different projects?

It uses git remote URLs or directory paths to isolate behaviors into project-specific containers, ensuring React patterns don't bleed into Python projects.

Is my code shared or uploaded to a cloud?

No, all observations and instincts are stored locally on your machine. The system only extracts patterns (instincts) and does not share raw code or conversation data.

How are instincts evolved into skills?

By using the /evolve command, the system identifies clusters of related high-confidence instincts and compiles them into structured skill files or automated commands.

What is an 'instinct' in this skill?

An instinct is a small, atomic learned behavior—such as a preference for a specific functional programming style—backed by observation evidence and a confidence score.

Continuous Learning V2 (Instinct System)

Name: Continuous Learning V2 (Instinct System)
Author: affaan-m

byaffaan-m

•

112,913

•

开发者工具

Automates the extraction of project-specific patterns and developer preferences into a reusable "instinct" library for Claude Code.

Continuous Learning V2 is a sophisticated knowledge extraction system designed to make Claude smarter over time by observing user interactions and tool usage. By capturing patterns—such as preferred code styles, common error fixes, or specific architectural decisions—and converting them into atomic "instincts" with confidence scores, it ensures Claude adapts to your specific workflow. Version 2.1 introduces project-level isolation to prevent cross-contamination between different tech stacks, while allowing high-confidence behaviors to be promoted to a global scope for universal application.

主要功能

01Deterministic observation hooks for 100% reliability in pattern capture during sessions

02Global promotion system for universal best practices like security and git workflows

03112,913 GitHub stars

04Evolution engine to cluster related instincts into full-fledged skills, commands, or agents

05Atomic instinct extraction with weighted confidence scoring from 0.3 to 0.9

06Smart project scoping to isolate codebase-specific conventions (e.g., React vs. Python)

使用场景

01Automatically capturing and enforcing team-specific coding styles or architectural patterns

02Reducing repetitive manual corrections by teaching Claude project-specific API usage

03Building a personalized library of AI behaviors that evolve and improve with every session

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add affaan-m/everything-claude-code continuous-learning-v2

For use in Claude.ai and ChatGPT

主要功能

01Deterministic observation hooks for 100% reliability in pattern capture during sessions

02Global promotion system for universal best practices like security and git workflows

03112,913 GitHub stars

04Evolution engine to cluster related instincts into full-fledged skills, commands, or agents

05Atomic instinct extraction with weighted confidence scoring from 0.3 to 0.9

06Smart project scoping to isolate codebase-specific conventions (e.g., React vs. Python)

使用场景

01Automatically capturing and enforcing team-specific coding styles or architectural patterns

02Reducing repetitive manual corrections by teaching Claude project-specific API usage

03Building a personalized library of AI behaviors that evolve and improve with every session