What makes a Learner skill different from a simple code snippet?

Code snippets are static templates, whereas Learner skills capture the underlying principles, reasoning, and decision-making logic required to solve specific classes of problems.

How does Claude access these skills in later sessions?

Each skill includes 'triggers'—specific error fragments or keywords. When those patterns reappear in a future conversation, Claude retrieves the skill to apply the saved principles.

Is the Learner skill part of a larger framework?

Yes, it is a core capability of the oh-my-claudecode plugin, which provides multi-agent orchestration and advanced execution modes for Claude Code.

Where are the learned skills stored?

By default, skills are saved in the .omc/skills/ directory within your project, making them easy to version control and share with your team via Git.

Does this extract generic programming advice?

No. The system is designed to reject generic or easily searchable information, focusing instead on non-obvious, codebase-specific insights that require real debugging effort to uncover.

Contextual Knowledge Learner

Name: Contextual Knowledge Learner
Author: Yeachan-Heo

byYeachan-Heo

•

2,142

学習とドキュメンテーション

Extracts and codifies non-obvious coding heuristics and project-specific principles from active conversations to improve future problem-solving.

概要

The Learner skill transforms Claude from a generic assistant into a project-aware expert by identifying and documenting deep insights discovered during debugging or development. Unlike simple note-taking, it focuses on high-level decision-making heuristics and 'hard-won' lessons that are specific to a codebase but difficult to find via search engines. This ensures that once a complex bug or undocumented architectural pattern is resolved, Claude remembers the underlying logic to prevent regressions and streamline subsequent tasks across the entire development lifecycle.

主な機能

Automated quality validation to filter out generic or googleable content
Trigger-based recognition patterns for automated retrieval
2,142 GitHub stars
Multi-level storage for project-specific or universal insights
Standardized Markdown output compatible with version control
Principle-based extraction of decision-making heuristics

ユースケース

Documenting architectural 'gotchas' discovered during deep-dive debugging sessions
Building a persistent, shared knowledge base of codebase-specific best practices
Capturing undocumented workarounds for internal legacy systems

概要

主な機能

Automated quality validation to filter out generic or googleable content
Trigger-based recognition patterns for automated retrieval
2,142 GitHub stars
Multi-level storage for project-specific or universal insights
Standardized Markdown output compatible with version control
Principle-based extraction of decision-making heuristics

ユースケース

Documenting architectural 'gotchas' discovered during deep-dive debugging sessions
Building a persistent, shared knowledge base of codebase-specific best practices
Capturing undocumented workarounds for internal legacy systems