Guides and resources for AI engineers: an agents course, five handbooks, a hands-on second-brain project and a vetted reading list.
JavaScript agent-skillsai-agentsai-engineeringclaudeclaude-codeclaude-skillsknowledge-graphknowledge-managementllmmcpobsidianpersonal-knowledge-basepkmsecond-brain
readme: resources as tables
undefined-uiea32a7a23669
diff --git a/README.md b/README.mdindex 2636994..2454bcf 100644--- a/README.md+++ b/README.md@@ -154,101 +154,67 @@ you keep the folder and everything in it.The source document, and the research behind the design decisions in this guide.-**[Karpathy's llm-wiki gist](https://gist.github.com/karpathy/442a6bf555914893e9891c11519de94f)**-(4 April 2026). The pattern everything here builds on. Short, written as an idea-file to paste into your own agent rather than as a spec. Read it first.--**[From Local to Global: A Graph RAG Approach to Query-Focused-Summarization](https://arxiv.org/abs/2404.16130)** (arXiv:2404.16130, Microsoft-Research). Why chunk retrieval fails on questions about a whole corpus, which is-exactly what people want from a second brain. Code:-[microsoft/graphrag](https://github.com/microsoft/graphrag).--**[HippoRAG: Neurobiologically Inspired Long-Term Memory for-LLMs](https://arxiv.org/abs/2405.14831)** (arXiv:2405.14831). Graph retrieval-with personalised PageRank for multi-hop questions. The closest published-analogue of an agent walking links outward from a page.--**[Lost in the Middle: How Language Models Use Long-Contexts](https://arxiv.org/abs/2307.03172)** (arXiv:2307.03172, TACL). The-empirical reason not to paste your whole vault into context, and the basis for-the [context budget](docs/07-retrieval/context-budget.md) rules here.--**[Andy Matuschak's notes](https://notes.andymatuschak.org)**. Evergreen notes,-written in public. Also the strongest argument against this whole approach: if-the writing is the thinking, delegating it means not doing it.--Plus **How to Take Smart Notes** (Ahrens) and **Building a Second Brain**-(Forte), both written before agents existed, which is what makes them worth-reading now. Full list with notes: [resources/reading.md](resources/reading.md)-and [resources/papers.md](resources/papers.md).--## Tools--**Core stack.** [Obsidian](https://obsidian.md) for storage,-[Web Clipper](https://obsidian.md/clipper) for capture,-[Claude Code](https://code.claude.com/docs/en/setup) for maintenance.--**Plugins.** [Dataview](https://github.com/blacksmithgu/obsidian-dataview) for-queries over frontmatter,-[Templater](https://github.com/SilentVoid13/Templater) for pages you write by-hand, and-[Local REST API](https://github.com/coddingtonbear/obsidian-local-rest-api) if-you want MCP access. That plugin now ships a built-in MCP server, so a separate-one is often unnecessary; see [MCP for Obsidian](docs/02-setup/mcp-obsidian.md).+| Source | What it gives you |+|---|---|+| **[Karpathy's llm-wiki gist](https://gist.github.com/karpathy/442a6bf555914893e9891c11519de94f)** (4 Apr 2026) | The pattern everything here builds on. Short, written as an idea file to paste into your agent rather than as a spec. Read it first. |+| **[From Local to Global: GraphRAG](https://arxiv.org/abs/2404.16130)** arXiv:2404.16130 | Why chunk retrieval fails on questions about a whole corpus, which is exactly what people want from a second brain. Code: [microsoft/graphrag](https://github.com/microsoft/graphrag). |+| **[HippoRAG](https://arxiv.org/abs/2405.14831)** arXiv:2405.14831 | Graph retrieval with personalised PageRank for multi-hop questions. The closest published analogue of an agent walking links outward. |+| **[Lost in the Middle](https://arxiv.org/abs/2307.03172)** arXiv:2307.03172, TACL | The empirical reason not to paste your whole vault into context, and the basis for the [context budget](docs/07-retrieval/context-budget.md) rules. |+| **[Andy Matuschak's notes](https://notes.andymatuschak.org)** | Evergreen notes, written in public. Also the strongest argument against this approach: if the writing is the thinking, delegating it means not doing it. |+| **How to Take Smart Notes** (Ahrens) | Zettelkasten. Atomic notes and dense linking still hold; the manual labour is what killed it for most people. |+| **Building a Second Brain** (Forte) | Where the term comes from, and PARA. Most of the book is about maintenance work an agent removes. |-**Capture and processing.** [yt-dlp](https://github.com/yt-dlp/yt-dlp),-[youtube-transcript-api](https://github.com/jdepoix/youtube-transcript-api),-[OCRmyPDF](https://github.com/ocrmypdf/OCRmyPDF),-[Zotero](https://www.zotero.org).+Full notes: [reading.md](resources/reading.md) and [papers.md](resources/papers.md).-**Graph and search.** [ripgrep](https://github.com/BurntSushi/ripgrep),-[NetworkX](https://networkx.org), [Kuzu](https://kuzudb.com),-[Gephi](https://gephi.org).+## Tools-Full list, including alternatives to Obsidian:-[resources/tools.md](resources/tools.md).+| Category | Tool | What it does |+|---|---|---|+| Core | [Obsidian](https://obsidian.md) | Local markdown editor. The vault is a plain folder |+| Core | [Web Clipper](https://obsidian.md/clipper) | Official extension, saves articles into `raw/` |+| Core | [Claude Code](https://code.claude.com/docs/en/setup) | The agent that maintains the wiki. Paid plan required |+| Plugin | [Dataview](https://github.com/blacksmithgu/obsidian-dataview) | Queries over frontmatter. Answers "which pages have property P" |+| Plugin | [Templater](https://github.com/SilentVoid13/Templater) | Templates for pages you write by hand |+| Plugin | [Local REST API](https://github.com/coddingtonbear/obsidian-local-rest-api) | Vault over a local API. Now ships a built-in MCP server, so a separate one is often unnecessary |+| MCP | [mcp-obsidian](https://github.com/MarkusPfundstein/mcp-obsidian) | The widely documented third-party server. See [MCP for Obsidian](docs/02-setup/mcp-obsidian.md) |+| Capture | [yt-dlp](https://github.com/yt-dlp/yt-dlp) | Subtitles and audio from video platforms |+| Capture | [youtube-transcript-api](https://github.com/jdepoix/youtube-transcript-api) | Transcripts from Python |+| Capture | [OCRmyPDF](https://github.com/ocrmypdf/OCRmyPDF) | Text layer for scanned PDFs |+| Capture | [Zotero](https://www.zotero.org) | Reference manager with markdown export |+| Graph | [ripgrep](https://github.com/BurntSushi/ripgrep) | Fast enough that most vaults never need an index |+| Graph | [NetworkX](https://networkx.org) | Graph analysis once you have exported an edge list |+| Graph | [Kuzu](https://kuzudb.com) | Embedded graph database, no server |+| Graph | [Gephi](https://gephi.org) | Visual graph exploration beyond Obsidian's view |++Full list, including alternatives to Obsidian: [tools.md](resources/tools.md).## Other implementationsWorth reading before writing your own page contracts.-- **[Astro-Han/karpathy-llm-wiki](https://github.com/Astro-Han/karpathy-llm-wiki)**.- The pattern as one installable skill for Claude Code, Cursor and Codex.- Its list of what the author deliberately did not build is the most useful part.-- **[micuintus/llm-wiki](https://github.com/micuintus/llm-wiki)**. Deliberately- minimal, convention-based rather than code-enforced. Good counterpoint.-- **[NicholasSpisak/second-brain](https://github.com/NicholasSpisak/second-brain)**.- Four skills installable through npm, close to the original gist.-- **[eugeniughelbur/obsidian-second-brain](https://github.com/eugeniughelbur/obsidian-second-brain)**.- Around 43 commands, works across Claude, Codex and Gemini.+| Repo | What it is |+|---|---|+| [Astro-Han/karpathy-llm-wiki](https://github.com/Astro-Han/karpathy-llm-wiki) | The pattern as one installable skill for Claude Code, Cursor and Codex. Its list of what the author deliberately did not build is the most useful part |+| [micuintus/llm-wiki](https://github.com/micuintus/llm-wiki) | Deliberately minimal, convention-based rather than code-enforced. Good counterpoint |+| [NicholasSpisak/second-brain](https://github.com/NicholasSpisak/second-brain) | Four skills installable through npm, close to the original gist |+| [eugeniughelbur/obsidian-second-brain](https://github.com/eugeniughelbur/obsidian-second-brain) | Around 43 commands, works across Claude, Codex and Gemini |-More context on each: [resources/skills.md](resources/skills.md).+More context on each: [skills.md](resources/skills.md).## Graph and publishing repos-Beyond implementations of the pattern itself.--**Graph views for Obsidian.** [Juggl](https://github.com/HEmile/juggl) for an-interactive workspace graph with typed edges,-[ExcaliBrain](https://github.com/zsviczian/excalibrain) for a structured-mind-map derived from your links and Dataview fields, and-[Breadcrumbs](https://github.com/michaelpporter/breadcrumbs) for typed links plus-trees, matrices and Mermaid views over them.--**GraphRAG.** [microsoft/graphrag](https://github.com/microsoft/graphrag) is the-reference implementation;-[nano-graphrag](https://github.com/gusye1234/nano-graphrag) is the same idea in-about 1,100 readable lines; [LightRAG](https://github.com/HKUDS/LightRAG) adds a-vector layer for cheaper incremental updates.-[Awesome-GraphRAG](https://github.com/DEEP-PolyU/Awesome-GraphRAG) covers the-rest of the landscape.--**Publishing.** [Quartz](https://github.com/jackyzha0/quartz) turns a vault into-a static site with backlinks, local graph and search.--Notes on each, and how to judge a repo in this ecosystem:-[resources/repositories.md](resources/repositories.md).+| Purpose | Repo | What it adds |+|---|---|---|+| Graph view | [Juggl](https://github.com/HEmile/juggl) | Interactive workspace graph on Cytoscape.js, typed edges, expand and pin nodes |+| Graph view | [ExcaliBrain](https://github.com/zsviczian/excalibrain) | Structured mind-map deriving five relationship types from links and Dataview fields |+| Graph view | [Breadcrumbs](https://github.com/michaelpporter/breadcrumbs) | Typed links plus trees, matrices, Mermaid and Canvas export |+| GraphRAG | [microsoft/graphrag](https://github.com/microsoft/graphrag) | The reference implementation of the paper |+| GraphRAG | [nano-graphrag](https://github.com/gusye1234/nano-graphrag) | The same idea in about 1,100 readable lines. Open this one to understand the pipeline |+| GraphRAG | [LightRAG](https://github.com/HKUDS/LightRAG) | Adds a vector layer for cheaper incremental updates |+| GraphRAG | [Awesome-GraphRAG](https://github.com/DEEP-PolyU/Awesome-GraphRAG) | Curated map of the rest of the landscape |+| Publishing | [Quartz](https://github.com/jackyzha0/quartz) | Vault to static site with backlinks, local graph and search |++Notes on each, and how to judge a repo in this ecosystem: [repositories.md](resources/repositories.md).## Contributing