MMCPMarkets
TC

ttguy0707/codojo

Developer tools

This page was translated automatically. Technical details may differ from the original. View the original

73 stars Quality 62 Trend 18MIT
Repository Install

Code Dojo: Let AI accompany you from not understanding projects to independently modifying them · AI-powered learning workflow for real code projects

Overview

It's not just a simple 'code interpretation' tool, but a learning protocol built around real projects: AI scans the project, then assesses your abilities, generates a personalized learning plan, and then guides you through theoretical understanding and code practice based on key points. All progress is stored in .codojo/, enabling cross-session continuation. - Students want to quickly get started with an open source project - Newcomers taking over existing company projects need to build a comprehensive project overview - Practitioners migrating the tech stack want to learn around real code - After completing the project, they hope to make a small feature modification to test understanding Codojo does not rely on the context of a single chat to preserve learning status; instead, it writes evaluation results, study plans, schedules, notes, and modification plans into the target project's .codojo/ directory. - Continue learning after starting a new session - let the AI reload the learning status - review what you learned, what you skipped, and where you got stuck - manage the learning process together with project code Each knowledge point must be tied to real project documents; after theoretical explanation, it moves on to practical tasks. Practice is not a fictional demo but a small, clear change within the target project. By default, the current directory is used as the target project. That is, npm link runs in the Codojo repository, and dojo install runs in the code repository you want to learn. It writes the current npm global bin directory into the shell profile (such as ~/.zshrc or ~/.bashrc). Regular dojo install does not modify shell configurations or output such repair prompts. dojo install is idempotent and can be executed repeatedly. It only updates the shared and dojo-skills managed by Codojo, does not delete other skills, nor does it delete the .codojo/ learning file.

Overview (machine translated)

Code Dojo: Let AI Accompany You from ‘Not Understanding Projects’ to ‘Able to Independently Modify Things’

Code Dojo · Code Dojo

English · Chinese Simplified

Quick Start · Learning Process · Document Protocol · Product Examples · Skills · CLI

Introduction

Codojo is a code dojo for Codex and Claude Code.

It’s not just a simple ‘code interpretation’ tool, but a learning protocol centered around real projects: AI first scans the project, then assesses your abilities, generates a personalized learning plan, and then guides you through theoretical understanding and code practice based on key points. All progress is accumulated to .codojo/, enabling cross-session continuation.

Suitable for these scenarios:

  • Students want to quickly get started with an open-source project
  • Newcomers taking over existing projects need to establish a comprehensive project overview
  • Practitioners migrating their tech stack to learn around real code
  • After completing the project, hope to make a small feature modification to test understanding

Core Philosophy

Documents are progress

Codojo does not rely on the context of a single chat to preserve learning status; instead, it writes evaluation results, study plans, schedules, notes, and modification plans into the target project’s .codojo/ directory.

This means you can:

  • Continue learning after starting a new session
  • Let AI reread the learning status
  • Review what you learned, what you skipped, and where you got stuck
  • Manage the learning process together with project code

Real Code Driver

Each knowledge point must be tied to real project documents; after theoretical explanation, it moves on to practical tasks. Practice is not a fictional demo but a small, clear change within the target project.

Users can choose to complete the practice or explicitly skip it; The AI will write the status back to the progress file.

Learning Process

Stage Skill What to do Output
S0 Project Scan dojo-init / dojo-stage Read the methodology, scan the project structure, tech stack, and current progress Route to the correct stage
S1 Competency Assessment dojo-assess Generating questionnaires based on the project technology stack, confirming user base question by item open-questions.md
S2 Plan Generation dojo-plan Generate personalized learning paths and progress charts task.md、schedule.md
S3 Formal Teaching dojo-teach Theoretical explanations and real code practice based on knowledge points Continuously updated schedule.md
S4 Project Mods dojo-hack After completing the coursework, propose directions for transformation and assist in implementing plan.md and code changes

Support Skills:

Skill Purpose
dojo-stage Stage router, automatically determines current progress

This page was translated automatically. Technical details may differ from the original.

Recommended Tools

Try a different keyword or remove a filter.

Install

npx skillfish add ttguy0707/codojo
New to MCP? Start with the mcpskill.org guide →