Does this skill share my data across different projects?

By default, version 2.1 uses project-scoping to keep sensitive project patterns isolated. You can manually 'promote' high-value instincts to a global scope if desired.

What is an 'instinct' in this system?

An instinct is an atomic, learned behavior (e.g., 'prefer functional components') derived from session observations with a specific confidence score and scope.

How does it learn without slowing down my performance?

It uses deterministic system hooks to log interactions, which are then analyzed by a background agent during idle time, ensuring zero latency during your active coding session.

How do instincts evolve into full skills?

The system uses the /evolve command to cluster related instincts together. Once a cluster reaches a high confidence threshold, it generates a structured skill or command file.

Intelligent Continuous Learning

Name: Intelligent Continuous Learning
Author: affaan-m

byaffaan-m

•

172,009

•

생산성 및 워크플로

Captures and evolves behavioral patterns into project-specific instincts and reusable skills through deterministic observation.

This skill transforms Claude Code into an adaptive development partner by monitoring session interactions via hooks and distilling them into 'instincts'—atomic, evidence-based behavioral patterns. Version 2.1 introduces sophisticated project-scoping, ensuring that project-specific patterns like React component structures stay within their respective environments while global best practices, such as input validation, are shared across all projects. By utilizing a background analysis agent and a confidence-scoring model, the system evolves these raw observations into production-ready skills, commands, and specialized agents, effectively automating the optimization of your local development workflow and reducing repetitive instructions.

주요 기능

01Deterministic Hook-based Observation: Captures 100% of tool usage for reliable pattern recognition.

02Confidence-Weighted Instincts: Assigns reliability scores (0.3-0.9) based on evidence frequency and user feedback.

03Behavioral Evolution: Automatically clusters related instincts to generate new skills, commands, and agents.

04Project-Aware Scoping: Automatically isolates project-specific coding styles using git remote detection.

05172,009 GitHub stars

06Background Analysis: Uses a lightweight agent to analyze logs and update instincts without interrupting the main workflow.

사용 사례

01Converting repetitive multi-step workflows into automated custom commands.

02Promoting high-confidence security and testing practices from one project to a global behavior set.

03Automating project-specific coding style enforcement without manual prompt engineering.

주요 기능

01Deterministic Hook-based Observation: Captures 100% of tool usage for reliable pattern recognition.

02Confidence-Weighted Instincts: Assigns reliability scores (0.3-0.9) based on evidence frequency and user feedback.

03Behavioral Evolution: Automatically clusters related instincts to generate new skills, commands, and agents.

04Project-Aware Scoping: Automatically isolates project-specific coding styles using git remote detection.

05172,009 GitHub stars

06Background Analysis: Uses a lightweight agent to analyze logs and update instincts without interrupting the main workflow.

사용 사례

01Converting repetitive multi-step workflows into automated custom commands.

02Promoting high-confidence security and testing practices from one project to a global behavior set.

03Automating project-specific coding style enforcement without manual prompt engineering.