What is the significance of the confidence score?

Confidence scores from 0.3 to 0.9 determine how strictly Claude follows a learned pattern, ranging from tentative suggestions (0.3) to core, auto-approved behaviors (0.9).

Can I share my learned instincts with my development team?

Absolutely. The skill includes /instinct-export and /instinct-import commands specifically designed to help teams synchronize coding standards and best practices discovered by the AI.

How does Continuous Learning v2 differ from standard Claude memory?

Unlike standard context-based memory, this skill uses deterministic hooks to capture 100% of interactions and organizes them into weighted 'instincts' that evolve into permanent, reusable skills.

Is my code and conversation data kept private?

Yes, all observations are stored locally on your machine. Only the abstract patterns (instincts) are ever visible to the analysis agent, and nothing is shared unless you manually export it.

Continuous Learning v2

Name: Continuous Learning v2
Author: unju-ai

byunju-ai

0•

학습 및 문서화

Transforms Claude Code sessions into a persistent knowledge base of personalized instincts and automated behavioral patterns.

Continuous Learning v2 is an advanced instinct-based architecture that enables Claude to learn from every interaction through deterministic hooks. By capturing tool usage and user corrections in real-time, it builds a library of atomic instincts with confidence scoring ranging from 0.3 to 0.9. These instincts gradually evolve into full-fledged skills, custom commands, and specialized agents, ensuring Claude adapts to your specific coding style, preferred libraries, and architectural patterns over time without the need for manual prompt engineering.

주요 기능

01Deterministic hook-based observation for 100% reliable session capture

02Background pattern detection for user corrections and recurring error resolutions

03Seamless import and export of learning patterns for team-wide sharing

04Atomic instinct modeling with weighted confidence scoring (0.3–0.9)

05Automated evolution from raw observations to structured skills and agents

060 GitHub stars

사용 사례

01Generating custom CLI commands and specialized agents based on historical project interactions

02Automatically adopting and enforcing team-specific functional programming or validation patterns

03Building a persistent memory of recurring bug fixes and project-specific debugging workflows

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add unju-ai/ecc continuous-learning-v2

For use in Claude.ai and ChatGPT