{"id":51551316,"url":"https://github.com/TakaGoto/rag-learning-academy","last_synced_at":"2026-07-12T14:01:17.440Z","repository":{"id":345964842,"uuid":"1188042366","full_name":"TakaGoto/rag-learning-academy","owner":"TakaGoto","description":"A structured, multi-agent Claude Code learning environment for mastering Retrieval-Augmented Generation (RAG)","archived":false,"fork":false,"pushed_at":"2026-04-03T15:49:32.000Z","size":509,"stargazers_count":2,"open_issues_count":0,"forks_count":0,"subscribers_count":0,"default_branch":"main","last_synced_at":"2026-04-10T08:59:39.042Z","etag":null,"topics":[],"latest_commit_sha":null,"homepage":null,"language":"Python","has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":"mit","status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/TakaGoto.png","metadata":{"files":{"readme":"README.md","changelog":null,"contributing":"CONTRIBUTING.md","funding":null,"license":"LICENSE","code_of_conduct":null,"threat_model":null,"audit":null,"citation":null,"codeowners":null,"security":null,"support":null,"governance":null,"roadmap":null,"authors":null,"dei":null,"publiccode":null,"codemeta":null,"zenodo":null,"notice":null,"maintainers":null,"copyright":null,"agents":"AGENTS.md","dco":null,"cla":null}},"created_at":"2026-03-21T14:40:36.000Z","updated_at":"2026-04-03T15:49:40.000Z","dependencies_parsed_at":null,"dependency_job_id":null,"html_url":"https://github.com/TakaGoto/rag-learning-academy","commit_stats":null,"previous_names":["takagoto/rag-learning-academy"],"tags_count":0,"template":false,"template_full_name":null,"purl":"pkg:github/TakaGoto/rag-learning-academy","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/TakaGoto%2Frag-learning-academy","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/TakaGoto%2Frag-learning-academy/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/TakaGoto%2Frag-learning-academy/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/TakaGoto%2Frag-learning-academy/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/TakaGoto","download_url":"https://codeload.github.com/TakaGoto/rag-learning-academy/tar.gz/refs/heads/main","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/TakaGoto%2Frag-learning-academy/sbom","scorecard":null,"host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":286080680,"owners_count":35393398,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2026-05-26T15:22:16.424Z","status":"online","status_checked_at":"2026-07-12T02:00:06.386Z","response_time":87,"last_error":null,"robots_txt_status":"success","robots_txt_updated_at":"2025-07-24T06:49:26.215Z","robots_txt_url":"https://github.com/robots.txt","online":true,"can_crawl_api":true,"host_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub","repositories_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories","repository_names_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repository_names","owners_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners"}},"keywords":[],"created_at":"2026-07-10T00:00:52.211Z","updated_at":"2026-07-12T14:01:17.433Z","avatar_url":"https://github.com/TakaGoto.png","language":"Python","funding_links":[],"categories":["Documentation, Knowledge \u0026 Learning"],"sub_categories":[],"readme":"# RAG Learning Academy\n\n[![License: MIT](https://img.shields.io/badge/License-MIT-blue.svg)](LICENSE)\n[![CI](https://github.com/TakaGoto/rag-learning-academy/actions/workflows/ci.yml/badge.svg)](https://github.com/TakaGoto/rag-learning-academy/actions/workflows/ci.yml)\n[![Python 3.10+](https://img.shields.io/badge/python-3.10+-green.svg)](https://www.python.org/downloads/)\n\nA structured, multi-agent learning environment for mastering Retrieval-Augmented Generation (RAG) — powered by Claude Code.\n\nInspired by [Claude Code Game Studios](https://github.com/Donchitos/Claude-Code-Game-Studios), reimagined as an interactive learning guide for RAG systems.\n\n\u003cp align=\"center\"\u003e\n  \u003cimg src=\"assets/demo.gif\" alt=\"RAG Learning Academy demo\" width=\"720\"\u003e\n\u003c/p\u003e\n\n## What Is This?\n\nRAG Learning Academy transforms Claude Code into a **personal RAG tutor** with 20 specialized AI agents, 22 interactive slash commands, and a 9-module curriculum that takes you from \"what is RAG?\" to production deployment.\n\nEvery concept is paired with a hands-on exercise. Every exercise is paired with evaluation metrics. You learn by building.\n\n## Quick Start\n\n```bash\ngit clone https://github.com/TakaGoto/rag-learning-academy.git\ncd rag-learning-academy\nclaude                         # Open Claude Code\n/start                         # Begin your learning journey\n```\n\nThat's it. No API keys, no `pip install`, no setup. Dependencies get installed when you need them (the first hands-on lesson will guide you).\n\n## What's Inside\n\n### 20 Specialist Agents\n\nAgents don't auto-route — they answer directly when asked, and Claude will suggest the right specialist when a question goes deep into their domain.\n\n| Tier | Count | Role | Model |\n|------|-------|------|-------|\n| Directors | 3 | Curriculum, Architecture, Research | opus |\n| Domain Leads | 5 | Embedding, Retrieval, Indexing, Evaluation, Integration | opus |\n| Specialists | 12 | Chunking, Vector DB, Reranking, Prompt Engineering, Hybrid Search, Document Parsing, Metadata, Query Analysis, Deployment, Evaluation Metrics, Graph RAG, Multimodal | sonnet |\n\n### 22 Slash Commands\n\n| Command | What It Does |\n|---------|-------------|\n| `/start` | Assess your level, pick a track, get a working pipeline |\n| `/lesson` | Start or continue a lesson (checkpoint quizzes between modules) |\n| `/quiz` | Test your understanding |\n| `/build` | Build a RAG component step by step |\n| `/evaluate` | Run metrics on your pipeline |\n| `/debug-rag` | Diagnose common RAG failures |\n| `/compare` | Compare two approaches with live side-by-side output diffs |\n| `/benchmark` | Benchmark pipeline performance |\n| `/architecture` | Design a RAG system for a use case |\n| `/paper-review` | Walk through a research paper |\n| `/code-review` | Get feedback on your RAG code |\n| `/glossary` | Look up RAG terminology |\n| `/challenge` | Take on a hands-on challenge |\n| `/explain` | Deep-dive into any concept (supports ELI5 mode) |\n| `/roadmap` | View progress, badges, streaks, and export GitHub badges |\n| `/triage` | Not sure where to go? Get routed to the right skill |\n| `/audit-content` | Check materials for outdated references |\n| `/recap` | Quick summary of what you covered last session |\n| `/sandbox` | Spin up a minimal RAG pipeline instantly |\n| `/break-it` | Learn by debugging intentionally broken pipelines |\n| `/fix` | Skip the teaching, diagnose and fix your pipeline |\n| `/journal` | Write notes about what clicked or confused you |\n\n### 9-Module Curriculum\n\nEach lesson is tagged `core` or `optional`. The core path (~8.5 hours) gets you to a working, evaluated RAG system. Optional lessons add depth when you're ready.\n\n| # | Module | Lessons | Core | Key Topics |\n|---|--------|---------|------|------------|\n| 1 | Foundations | 4 | 4 | What is RAG, architecture, RAG vs fine-tuning |\n| 2 | Document Processing | 5 | 2 | Parsing, chunking strategies, metadata |\n| 3 | Embeddings | 5 | 2 | Models, vector spaces, similarity, fine-tuning |\n| 4 | Vector Databases | 5 | 2 | Chroma, Pinecone, pgvector, indexing |\n| 5 | Retrieval Strategies | 5 | 1 | Dense, sparse, hybrid, reranking, MMR |\n| 6 | Generation | 5 | 2 | Prompt engineering, grounding, citations |\n| 7 | Evaluation | 5 | 2 | RAGAS, retrieval/generation metrics |\n| 8 | Advanced Patterns | 5 | 0 | Agentic RAG, Graph RAG, CRAG, multimodal |\n| 9 | Production | 5 | 0 | Deployment, caching, monitoring, scaling |\n\n**Core path: ~8.5 hours** | Full curriculum: ~31 hours\n\n### 8 Milestones + 4 Proficiency Levels\n\nThe curriculum is broken into milestones — concrete checkpoints that mark real capability, not just lessons read.\n\n| # | Milestone | You Can Now... |\n|---|-----------|---------------|\n| 1 | First Light | Build a working RAG system from scratch |\n| 2 | Data Wrangler | Turn any document into retrieval-ready chunks |\n| 3 | Vector Navigator | Store and search embeddings effectively |\n| 4 | Retrieval Engineer | Find the right information for any query |\n| 5 | Prompt Architect | Generate grounded, cited answers |\n| 6 | Quality Guardian | Measure everything and improve with data |\n| 7 | Pattern Master | Go beyond basic RAG when it's warranted |\n| 8 | Production Ready | Deploy, monitor, and scale a RAG system |\n\nComplete milestones to earn proficiency levels:\n\n| Level | Milestones | What It Means |\n|-------|-----------|---------------|\n| **RAG Explorer** | 1-2 | Can build a basic pipeline and process documents |\n| **RAG Practitioner** | 3-5 | Can design retrieval systems with proper search and prompting |\n| **RAG Engineer** | 6-7 | Can evaluate, optimize, and apply advanced patterns |\n| **RAG Architect** | 8 + bonus | Can deploy, scale, and maintain production systems |\n\nSee [milestones.md](.claude/docs/reference/milestones.md) for full requirements and bonus milestones.\n\n### 3 Guided Projects\n\n| Level | Project | What You Build |\n|-------|---------|---------------|\n| Starter | Simple Q\u0026A | Basic RAG pipeline over documents |\n| Intermediate | Multi-Source Hybrid | Hybrid search + reranking + citations |\n| Advanced | Agentic RAG | Self-correcting RAG with routing and tools |\n\n### Progress Dashboard\n\nRun `make dashboard` to generate an HTML progress page showing your milestones, proficiency level, module completion, and quiz scores.\n\n## Tech Stack\n\nNo API keys required to start. The defaults work out of the box:\n\n| Component | Default (zero config) | Optional Upgrade |\n|-----------|----------------------|-----------------|\n| LLM | Claude Code (you're already running it) | Ollama (local, needs 8-16GB RAM) |\n| Embeddings | all-MiniLM-L6-v2 (local, no key) | OpenAI text-embedding-3-small |\n| Vector DB | ChromaDB (local, no setup) | Pinecone, pgvector, Qdrant |\n| Framework | LangChain | LlamaIndex |\n| Evaluation | RAGAS | Custom metrics |\n| Docs | PyPDF, BeautifulSoup | Unstructured, pdfplumber |\n\n\u003e **Running models locally?** Ollama is free but needs 8-16GB RAM for LLMs. Close heavy apps before running. If your machine slows down, switch to the API path. Local embedding models (all-MiniLM-L6-v2) are lightweight and run fine on most machines.\n\n## Project Structure\n\n```\nrag-learning-academy/\n├── CLAUDE.md                    # Master config, voice \u0026 tone, agent behavior\n├── .claude/\n│   ├── settings.json            # Hooks, permissions\n│   ├── agents/                  # 20 specialist agents\n│   ├── skills/                  # 22 slash commands\n│   ├── hooks/                   # Freshness checks, validation scripts\n│   ├── rules/                   # Path-scoped coding standards\n│   └── docs/\n│       ├── curriculum/          # 9-module learning path (core/optional tagged)\n│       ├── reference/           # Glossary, roster, milestones, standards\n│       └── templates/           # Architecture, eval, project templates\n├── src/                         # Your RAG code goes here\n├── scripts/                     # Dashboard generator, utilities\n├── projects/                    # Guided build projects\n├── tests/                       # 616 structural + content tests\n├── data/                        # Sample documents (6 files incl. 1,247-line chunking doc)\n└── progress/                    # Your learning progress + dashboard\n```\n\n## Learning Tracks\n\nRun `/start` and you'll be assessed into one of three tracks:\n\n| Track | Score | Modules | Level Earned | Est. Hours |\n|-------|-------|---------|-------------|------------|\n| Beginner | 0-3 | 1 → 4 | RAG Explorer | 15-25 |\n| Intermediate | 4-6 | 3 → 7 | RAG Practitioner | 25-40 |\n| Advanced | 7-10 | 6 → 9 | RAG Engineer | 30-45 |\n\nNot sure where to go? Run `/triage` to get routed based on your situation.\n\n## Content Freshness\n\nAcademy materials are actively monitored so nothing goes stale:\n\n- **Weekly CI** — checks PyPI versions, deprecated patterns, MTEB model health, review cycles ([details](.github/workflows/weekly-knowledge-check.yml))\n- **Monthly CI** — content age report, creates GitHub issues for stale files\n- **On-demand** — run `/audit-content` for a deep review with web search verification\n- **Frontmatter** — every content file has `last_reviewed`, `review_cycle`, and `staleness_risk` metadata\n\n## Philosophy\n\n\u003e \"Understand → Build → Evaluate → Iterate\"\n\n- Agents **teach**, they don't just code for you\n- Every concept comes with a **hands-on exercise**\n- Every exercise has **evaluation criteria**\n- You choose your path — agents advise, you decide\n- The tone is conversational and direct — like learning from a smart friend, not reading a textbook\n\n## Contributing\n\nContributions welcome! See [CONTRIBUTING.md](CONTRIBUTING.md) for guidelines on:\n\n- Reporting bugs and requesting features\n- Development setup and running tests (`make ci`)\n- Pull request process and checklist\n- Content guidelines and voice \u0026 tone\n\n```bash\nmake ci    # Run lint + shellcheck + 616 tests before submitting a PR\n```\n\n## Credits\n\nArchitecture inspired by [Claude Code Game Studios](https://github.com/Donchitos/Claude-Code-Game-Studios) by Donchitos.\n\n## License\n\n[MIT](LICENSE)\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2FTakaGoto%2Frag-learning-academy","html_url":"https://awesome.ecosyste.ms/projects/github.com%2FTakaGoto%2Frag-learning-academy","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2FTakaGoto%2Frag-learning-academy/lists"}