Where are the captured learnings stored?

Learnings are stored as Markdown files within your repository, typically in .claude/LEARNINGS.md or specialized reference directories like ci-doctor/ or lint-fixer/.

What kind of information does Opi consider 'a learning'?

Opi filters for non-obvious, actionable, and recurring patterns. It skips simple typos or one-off configuration issues in favor of architectural or logic patterns that Claude wouldn't know without being told.

How do I trigger the learning extraction process?

You can trigger it by typing '/opi', 'opi', or asking Claude 'What did we learn?' or 'Extract learnings' at any point during or at the end of a session.

Can I use Opi for non-English projects?

Yes, Opi includes triggers for multiple languages (such as 'oppia' and 'mitä opittiin' in Finnish) and can analyze code changes regardless of the primary language.

Does Opi automatically modify my documentation?

No, Opi identifies potential learnings and presents them for your review. It will only write the approved entries to your files after you give explicit consent.

Opi Learning Extractor

Name: Opi Learning Extractor
Author: Spectaculous-Code

bySpectaculous-Code

学習とドキュメンテーション

Extracts actionable insights from git commits and chat history to build a persistent project knowledge base.

概要

Opi is a specialized workflow skill designed to facilitate continuous improvement by capturing hard-earned knowledge during development sessions. By analyzing recent git activity—specifically bug fixes and refactors—alongside the current conversation context, it identifies non-obvious patterns, recurring mistakes, and project-specific solutions. It automates the process of updating project-specific learning files, ensuring that insights from one session are preserved and applied to future tasks, effectively reducing technical debt and preventing repetitive errors in AI-assisted coding.

主な機能

Categorizes learnings into domains like CI/CD, testing, or UI patterns
0 GitHub stars
Updates persistent .claude/LEARNINGS.md and domain-specific reference files
Identifies user corrections and non-obvious solutions from conversation history
Analyzes recent git logs for fix, refactor, and bugfix patterns automatically
Generates structured learning proposals with 'Wrong' vs 'Right' code examples

ユースケース

Automating end-of-session reflection to capture project-specific architectural decisions
Sharing technical context across different AI sessions by maintaining a project memory
Documenting recurring bug patterns to prevent regressions in CI/CD or frontend code

概要

主な機能

Categorizes learnings into domains like CI/CD, testing, or UI patterns
0 GitHub stars
Updates persistent .claude/LEARNINGS.md and domain-specific reference files
Identifies user corrections and non-obvious solutions from conversation history
Analyzes recent git logs for fix, refactor, and bugfix patterns automatically
Generates structured learning proposals with 'Wrong' vs 'Right' code examples

ユースケース

Automating end-of-session reflection to capture project-specific architectural decisions
Sharing technical context across different AI sessions by maintaining a project memory
Documenting recurring bug patterns to prevent regressions in CI/CD or frontend code