How does the skill categorize new information?

The skill analyzes your input for specific keywords to automatically assign categories like 'preference', 'convention', 'mistake', or 'decision' for better organization.

Are these learnings available in every session?

Yes, the Learn skill is integrated with a load-kb hook that automatically injects your saved knowledge base into the AI's context at the beginning of every session.

Can I remove a learning if it becomes outdated?

Yes, you can manage your project memory by manually editing the Markdown file or using the companion '/team-shinchan:forget' command.

Where are the learnings stored?

All captured information is saved locally in your project directory within the .shinchan-docs/learnings.md file, making it version-controllable and project-specific.

Does this skill share my project data externally?

No, all learnings are stored locally within your repository's documentation folder and are only used to provide context to your Claude Code instance.

Knowledge Base & Learning

Name: Knowledge Base & Learning
Author: seokan-jeong

byseokan-jeong

•

Learning & Documentation

Captures and categorizes project-specific patterns, rules, and preferences to build a persistent memory for AI agents.

The Learn skill enables users to explicitly teach Team-Shinchan agents crucial project context that might otherwise be missed. By manually inputting coding conventions, architectural decisions, or personal developer preferences, users create a structured knowledge base stored directly within the project's repository. This ensures that the AI agents maintain high-confidence context across multiple sessions, automatically loading these 'learned' insights at every startup to improve accuracy and reduce repetitive instructions.

Key Features

01Persistent project-specific storage in .shinchan-docs/learnings.md

02Automatic session-start loading via integrated knowledge base hooks

036 GitHub stars

04Automated categorization of insights into preferences, patterns, and conventions

05Manual knowledge capture using simple natural language commands

06Metadata enrichment with timestamps, confidence scores, and auto-generated tags

Use Cases

01Enforcing specific project coding standards like TypeScript strict mode or functional component usage

02Recording architectural decisions to ensure the AI suggests compatible implementation patterns

03Documenting previous mistakes or anti-patterns to avoid during the development lifecycle

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add seokan-jeong/team-shinchan learn

For use in Claude.ai and ChatGPT

Download Skill