continuous-learning-v2

Capture coding patterns and workflows from Claude Code sessions into confidence-weighted instincts.

Updated Mar 26, 2026
One-click install
npx skills add https://github.com/luongldptit/move-ticket --skill continuous-learning-v2-luongldptit
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: continuous-learning-v2
Source: https://github.com/luongldptit/move-ticket/tree/main/.agent/skills/continuous-learning-v2
Command: npx skills add https://github.com/luongldptit/move-ticket --skill continuous-learning-v2-luongldptit

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires python3, git, claude, and includes scripts (resource) components.

What problem does it solve?

This Skill solves the problem of losing learned coding patterns, workflows, and project conventions between Claude Code sessions, while eliminating the risk of cross-project pattern contamination that occurs when global rules apply universally across all codebases.

Core Features & Use Cases

  • Automatic Session Observation: Hooks capture 100% of tool use and user corrections from Claude Code sessions to detect patterns without manual input.
  • Project-Scoped Instincts: React, Python, or other project-specific patterns stay isolated to their respective repositories, while universal best practices (like input validation) are shared globally.
  • Confidence-Weighted Learning: Learned behaviors are assigned confidence scores from 0.3 (tentative) to 0.9 (near-certain) that evolve based on repeated observation and user feedback.
  • Instinct Evolution: Clustered related instincts can be evolved into full reusable skills, custom commands, or specialized agents.
  • Use Case: A developer working on a React e-commerce project and a Python data pipeline can have React hook patterns and Django REST conventions stay in their respective projects, while universal patterns like "always validate user input" are automatically shared across all projects.

Quick Start

Use the continuous-learning-v2 skill to enable automatic observation of your Claude Code sessions and start capturing project-specific coding patterns and workflows for reuse.

Frequently Asked Questions about continuous-learning-v2

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

FAQPage Schema
How do I retain learned coding patterns and workflows between Claude Code sessions?

To retain learned coding patterns between Claude Code sessions, you can use an automatic observation system that captures tool use and user corrections to preserve project-specific conventions for future reuse. This prevents losing workflows across sessions.

How do I prevent cross-project pattern contamination when sharing global coding rules?

To prevent cross-project pattern contamination, you can isolate project-specific patterns like React component structures from global best practices like input validation. This ensures project conventions stay scoped to their respective repositories.

Does the continuous learning instinct system require manual input to detect coding patterns?

No, the continuous learning instinct system does not require manual input. It uses automatic session observation hooks to capture 100% of tool use and user corrections to detect patterns backgroundly without manual intervention.

Can I use Claude Code to automatically promote universal project patterns to a global scope?

Yes, you can use Claude Code to automatically promote universal project patterns to a global scope. The system implements automatic promotion of universal patterns like grep-before-edit workflows to share them across all projects.

Do I need Python and Git to run project-scoped instinct detection for Claude Code?

Yes, you need Python3, Git, and Claude installed to run project-scoped instinct detection. These dependencies are required to execute the scripts that capture and preserve learned coding patterns from your sessions.

How do clustered instincts evolve into reusable skills, commands, and agents?

Clustered instincts evolve into reusable skills, commands, and agents through an instinct evolution system. Related learned behaviors with high confidence scores are automatically clustered and evolved into full reusable components for future automation.