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
resources: full catalog with stars and install counts
undefined-ui0f7eefcd1637
diff --git a/README.md b/README.mdindex 95b7203..f0fe215 100644--- a/README.md+++ b/README.md@@ -166,59 +166,75 @@ The source document, and the research behind the design decisions in this guide.Full notes: [reading.md](resources/reading.md) and [papers.md](resources/papers.md).-## Tools+## Tools and plugins-| 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 implementations--Worth reading before writing your own page contracts.--| 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: [skills.md](resources/skills.md).+Obsidian plugins ranked by installs from the official community stats, September+2026. The full catalog is in [plugins.md](resources/plugins.md) and+[tools.md](resources/tools.md).-## Graph and publishing repos--| Purpose | Repo | What it adds |+| Purpose | Pick | Installs |+|---|---|---|+| Agent in the editor | [Claudian](https://github.com/yishentu/claudian) | 2.0M |+| Agent in the editor | [Copilot](https://github.com/logancyang/obsidian-copilot) | 1.8M |+| Suggests links | [Smart Connections](https://github.com/brianpetro/obsidian-smart-connections) | 1.2M |+| MCP access | [Local REST API with MCP](https://github.com/coddingtonbear/obsidian-local-rest-api) | 712K |+| Queries over frontmatter | [Dataview](https://github.com/blacksmithgu/obsidian-dataview) | 4.9M |+| Templates | [Templater](https://github.com/SilentVoid13/Templater) | 5.5M |+| Version control | [Git](https://github.com/Vinzent03/obsidian-git) | 3.1M |+| Migrating in | [Importer](https://github.com/obsidianmd/obsidian-importer) | 1.6M |+| Broken links and orphans | [Find unlinked files](https://github.com/Vinzent03/find-unlinked-files) | 225K |+| Flashcards from notes | [Spaced Repetition](https://github.com/st3v3nmw/obsidian-spaced-repetition) | 592K |+| Structured mind-map | [ExcaliBrain](https://github.com/zsviczian/excalibrain) | 336K |+| Interactive graph | [Juggl](https://github.com/HEmile/juggl) | 136K |++Outside Obsidian: [Web Clipper](https://obsidian.md/clipper) for capture,+[Claude Code](https://code.claude.com/docs/en/setup) for maintenance,+[yt-dlp](https://github.com/yt-dlp/yt-dlp) and+[OCRmyPDF](https://github.com/ocrmypdf/OCRmyPDF) for processing,+[ripgrep](https://github.com/BurntSushi/ripgrep),+[NetworkX](https://networkx.org), [Kuzu](https://kuzudb.com) and+[Gephi](https://gephi.org) for the graph, and+[Quartz](https://github.com/jackyzha0/quartz) to publish.++## Skills and other implementations++Counted individually: a repo shipping sixteen skills counts as sixteen. Stars+from the GitHub API, September 2026.++| Repo | Stars | Ships |+|---|---|---|+| [AgriciDaniel/claude-obsidian](https://github.com/AgriciDaniel/claude-obsidian) | 14.7K | 16 skills, 3 subagents, role presets |+| [Astro-Han/karpathy-llm-wiki](https://github.com/Astro-Han/karpathy-llm-wiki) | 2.2K | 1 skill covering ingest, compile, query, lint |+| [ballred/obsidian-claude-pkm](https://github.com/ballred/obsidian-claude-pkm) | 1.9K | 13 skills, 4 subagents, full starter kit |+| [coleam00/second-brain-starter](https://github.com/coleam00/second-brain-starter) | 768 | 1 skill that interviews you first |+| [NicholasSpisak/second-brain](https://github.com/NicholasSpisak/second-brain) | 704 | 4 skills, npm installer |+| [eugeniughelbur/obsidian-second-brain](https://github.com/eugeniughelbur/obsidian-second-brain) | small | 47 commands, works across three agents |+| [micuintus/llm-wiki](https://github.com/micuintus/llm-wiki) | small | 1 skill, deliberately minimal |++Where the format itself is defined:+[anthropics/skills](https://github.com/anthropics/skills) (20 skills),+[obra/superpowers](https://github.com/obra/superpowers) (14),+[VoltAgent/awesome-agent-skills](https://github.com/VoltAgent/awesome-agent-skills)+(index of 1,000+). Notes on each: [skills.md](resources/skills.md).++## Graph, RAG and memory repos++| Purpose | Repo | Stars ||---|---|---|-| 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 |-| Graph view | [Graph Analysis](https://github.com/SkepticMystic/graph-analysis) | Graph algorithms and similarity measures over your vault, surfaces unlinked connections |-| Graph view | [3D Graph](https://github.com/AlexW00/obsidian-3d-graph) | The vault as a 3D force graph. Shows cluster structure a 2D hairball hides |-| Linking | [Smart Connections](https://github.com/brianpetro/obsidian-smart-connections) | Local embedding model suggests related notes while you write. No API key |-| Build a graph | [Graphify](https://github.com/Graphify-Labs/graphify) | `/graphify` skill for Claude Code and others. Turns a folder of code, docs, PDFs and screenshots into a queryable graph. Every edge labelled extracted or inferred |-| 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).+| Build a graph from any folder | [Graphify](https://github.com/Graphify-Labs/graphify) | 116K |+| Graph RAG, incremental | [LightRAG](https://github.com/HKUDS/LightRAG) | 39K |+| Graph RAG, reference | [microsoft/graphrag](https://github.com/microsoft/graphrag) | 36K |+| Graph RAG, readable | [nano-graphrag](https://github.com/gusye1234/nano-graphrag) | 4.0K |+| Multi-hop retrieval | [HippoRAG](https://github.com/OSU-NLP-Group/HippoRAG) | 4.0K |+| The landscape | [Awesome-GraphRAG](https://github.com/DEEP-PolyU/Awesome-GraphRAG) | 2.6K |+| Agent memory | [mem0](https://github.com/mem0ai/mem0) | 65K |+| Temporal knowledge graphs | [graphiti](https://github.com/getzep/graphiti) | 31K |+| Graph plus vector memory | [cognee](https://github.com/topoteretes/cognee) | 31K |+| MCP server index | [awesome-mcp-servers](https://github.com/punkpeye/awesome-mcp-servers) | 95K |++Alternative homes for a vault, from Logseq to AFFiNE, plus RAG frameworks and+AI-native note apps: [repositories.md](resources/repositories.md) and+[tools.md](resources/tools.md).## Contributingdiff --git a/docs/02-setup/claude-code-setup.md b/docs/02-setup/claude-code-setup.mdindex 7e006e3..09266ec 100644--- a/docs/02-setup/claude-code-setup.md+++ b/docs/02-setup/claude-code-setup.md@@ -26,6 +26,21 @@ and retrieval sections of this guide.Installation instructions for each platform are in the [official setupdocs](https://code.claude.com/docs/en/setup).+## Option C: inside Obsidian++Community plugins now embed the agent in the editor, and they are heavily used:+[Claudian](https://github.com/yishentu/claudian) sits above two million installs+and [Copilot](https://github.com/logancyang/obsidian-copilot) is close behind.+Both run Claude Code, Codex or a local agent in a pane next to your notes.++Worth knowing about because it removes the window switching entirely. Worth+being careful with for the same reason this guide keeps ingestion on a+schedule: an agent in the editor invites ad-hoc edits that never make it into+`log.md`, and an unlogged vault is one you cannot audit.++A reasonable split is the plugin for conversation and drafting, the terminal or+a scheduled task for anything that writes wiki pages.+## Plan requirementClaude Code needs a paid account: Pro, Max, Team, Enterprise, or a Consolediff --git a/docs/02-setup/obsidian-plugins.md b/docs/02-setup/obsidian-plugins.mdindex dd6d1ce..1a1dab1 100644--- a/docs/02-setup/obsidian-plugins.md+++ b/docs/02-setup/obsidian-plugins.md@@ -33,6 +33,23 @@ Templates with variables and dates, for the page shapes innotes you write by hand. The agent does not need it, it writes frontmatterdirectly.+## Worth adding later++- **[Git](https://github.com/Vinzent03/obsidian-git)**, the third most installed+ plugin in the catalog. Commits from inside Obsidian, which pairs with+ [versioning](../09-maintenance/versioning-with-git.md).+- **[Importer](https://github.com/obsidianmd/obsidian-importer)**, official,+ pulls in Notion, Evernote, Roam, Bear and Apple Notes. The first step if you+ are migrating an existing pile.+- **[Find orphaned files and broken links](https://github.com/Vinzent03/find-unlinked-files)**,+ the same checks as [lint](../09-maintenance/lint-and-health.md) without leaving+ the app.+- **[Omnisearch](https://github.com/scambier/obsidian-omnisearch)** for better+ full-text search, including inside PDFs.++Full ranked list with install counts:+[resources/plugins.md](../../resources/plugins.md).+## What to avoidPlugins that rewrite files in the background. Anything that auto-formats,diff --git a/docs/08-outputs/teaching-yourself.md b/docs/08-outputs/teaching-yourself.mdindex 705dd7f..2fe3b32 100644--- a/docs/08-outputs/teaching-yourself.md+++ b/docs/08-outputs/teaching-yourself.md@@ -22,9 +22,9 @@ The reason spaced repetition has low adoption is that writing cards is tedious.The vault removes that step: concept pages are the source material, and the agentcan generate cards from them.-The Obsidian spaced repetition plugins read cards written in your notes, so the-cards live alongside the concept rather than in a separate app that drifts out of-sync.+The [Spaced Repetition](https://github.com/st3v3nmw/obsidian-spaced-repetition)+plugin reads cards written inside your notes, so they live alongside the concept+rather than in a separate app that drifts out of sync.Keep it to a small number of pages that actually matter. Reviewing everything youever saved is a job, not a habit.diff --git a/docs/09-maintenance/lint-and-health.md b/docs/09-maintenance/lint-and-health.mdindex 7fd6e69..b2964fd 100644--- a/docs/09-maintenance/lint-and-health.md+++ b/docs/09-maintenance/lint-and-health.md@@ -41,6 +41,10 @@ ingest, where it costs more than it catches.Run it on a clean git state, so a wrong repair is one `git checkout` away.+Inside Obsidian, [Find orphaned files and broken+links](https://github.com/Vinzent03/find-unlinked-files) covers the same two+checks visually.+For a fast read without invoking the agent:```bashdiff --git a/docs/10-troubleshooting/faq.md b/docs/10-troubleshooting/faq.mdindex 5b99b54..08f8c98 100644--- a/docs/10-troubleshooting/faq.md+++ b/docs/10-troubleshooting/faq.md@@ -35,7 +35,9 @@ Logseq, Foam, or a plain editor all work.**What if I already have a Notion or Evernote full of notes?**-Export to markdown, drop it in `raw/`, and backfill oldest first. Expect the+Use the official [Importer](https://github.com/obsidianmd/obsidian-importer)+plugin, which handles Notion, Evernote, Roam, Bear and Apple Notes. Then treat+the result as raw material: backfill oldest first. Expect theexport to be messy. Fix extraction before ingesting rather than hoping the agentcopes.diff --git a/resources/README.md b/resources/README.mdindex cf876c6..f578392 100644--- a/resources/README.md+++ b/resources/README.md@@ -1,14 +1,28 @@# Resources-Curated, not exhaustive. Every link here was opened and checked, and anything-that exists only in secondary coverage is marked as such.--- [Tools](tools.md). Editors, capture, agents, graph and search-- [Skills and implementations](skills.md). Agent skills and open-source builds- of this pattern-- [Repositories](repositories.md). Graph plugins, GraphRAG implementations,- publishing+A catalog, not a link dump. Every entry was checked, and every number comes from+the source itself: GitHub stars from the API, Obsidian plugin installs from the+official community stats file that the app itself is built on.++Figures are from **September 2026** and will drift. The relative picture holds+much longer than the absolute numbers.++- [Plugins](plugins.md). Obsidian plugins, ranked by installs+- [Tools](tools.md). Editors, capture, search, graph, everything outside Obsidian+- [Skills and agents](skills.md). Agent skills for this pattern, counted individually+- [Repositories](repositories.md). Graph, RAG, memory and PKM repos by stars- [Reading](reading.md). The ideas behind the pattern- [Papers](papers.md). Retrieval, graphs and long-context research-If something here has gone stale or a better option exists, open a pull request.+## How entries are chosen++**Repositories** need real adoption, roughly a thousand stars, unless they are+small on purpose and worth reading anyway, in which case that is stated.++**Plugins** are ranked by installs rather than stars. Installs measure use;+stars measure attention, and the two diverge sharply in this ecosystem.++**Skills are counted individually.** A repo shipping sixteen skills counts as+sixteen, because that is what you actually get.++Anything I could not verify is named without a link rather than guessed at.diff --git a/resources/plugins.md b/resources/plugins.mdnew file mode 100644index 0000000..1ebe32f--- /dev/null+++ b/resources/plugins.md@@ -0,0 +1,69 @@+# Obsidian plugins++Ranked by installs from Obsidian's own community stats, September 2026. The+catalog holds over 7,400 plugins; these are the ones that matter for an+agent-maintained vault.++## Agents inside Obsidian++A route this guide's [setup](../docs/02-setup/claude-code-setup.md) originally+skipped: instead of running the agent in a terminal pointed at the vault folder,+run it inside Obsidian.++| Plugin | Installs | What it does |+|---|---|---|+| [Claudian](https://github.com/yishentu/claudian) | 2,035,966 | Embeds Claude Code, Codex and other local agents as collaborators in the vault |+| [Copilot](https://github.com/logancyang/obsidian-copilot) | 1,832,464 | Runs Claude Code, Codex and OpenCode inside your vault |+| [Smart Connections](https://github.com/brianpetro/obsidian-smart-connections) | 1,189,513 | Local embedding model surfaces related notes while you write. No API key |+| [Local REST API with MCP](https://github.com/coddingtonbear/obsidian-local-rest-api) | 712,443 | Vault over a secure local API, now with a built-in MCP server |+| [Agent Client](https://github.com/rait-09/obsidian-agent-client) | 261,371 | Chat with Claude Code, Codex and Gemini CLI over the Agent Client Protocol |+| [Smart Composer](https://github.com/glowingjade/obsidian-smart-composer) | 171,863 | AI chat with note context and one-click edits |++## Structure and queries++| Plugin | Installs | What it does |+|---|---|---|+| [Templater](https://github.com/SilentVoid13/Templater) | 5,527,936 | Dynamic templates. For the pages you write by hand |+| [Dataview](https://github.com/blacksmithgu/obsidian-dataview) | 4,918,446 | Query language over your frontmatter. Answers "which pages have property P" |+| [Metadata Menu](https://github.com/mdelobelle/metadatamenu) | 330,430 | Manage frontmatter fields at scale. Useful once your [schema](../docs/04-structuring/frontmatter-schema.md) is fixed |+| [Supercharged Links](https://github.com/mdelobelle/obsidian_supercharged_links) | 210,926 | Styles links by the target note's frontmatter, so page type is visible inline |++## Graph++| Plugin | Installs | What it does |+|---|---|---|+| [ExcaliBrain](https://github.com/zsviczian/excalibrain) | 336,025 | Structured mind-map deriving five relationship types from links, Dataview fields and tags |+| [Juggl](https://github.com/HEmile/juggl) | 135,614 | Interactive Cytoscape.js graph with typed edges and a saveable workspace mode |+| [Graph Analysis](https://github.com/SkepticMystic/graph-analysis) | community plugin | Graph algorithms and similarity measures over the vault, surfaces unlinked connections |+| [Breadcrumbs](https://github.com/michaelpporter/breadcrumbs) | community plugin | Typed links plus trees, matrices, Mermaid and Canvas export |+| [3D Graph](https://github.com/AlexW00/obsidian-3d-graph) | community plugin | The vault as a 3D force graph. Shows cluster structure a 2D hairball hides |++## Maintenance++| Plugin | Installs | What it does |+|---|---|---|+| [Git](https://github.com/Vinzent03/obsidian-git) | 3,110,571 | Commit and sync the vault from inside Obsidian. See [versioning](../docs/09-maintenance/versioning-with-git.md) |+| [Omnisearch](https://github.com/scambier/obsidian-omnisearch) | 1,837,484 | Better full-text search, including inside PDFs |+| [Importer](https://github.com/obsidianmd/obsidian-importer) | 1,638,001 | Official importer from Notion, Evernote, Roam, Bear, Apple Notes. The first step for anyone [backfilling](../docs/03-ingestion/bulk-backfill.md) |+| [Linter](https://github.com/platers/obsidian-linter) | 1,044,521 | Formats notes to consistent rules. Set it carefully, it rewrites files |+| [Find orphaned files and broken links](https://github.com/Vinzent03/find-unlinked-files) | 224,687 | Exactly what [lint](../docs/09-maintenance/lint-and-health.md) checks, without leaving Obsidian |++## Capture and study++| Plugin | Installs | What it does |+|---|---|---|+| [Excalidraw](https://github.com/zsviczian/obsidian-excalidraw-plugin) | 7,795,993 | The most installed plugin in the catalog. Diagrams that live in the vault as files |+| [QuickAdd](https://github.com/chhoumann/quickadd) | 2,082,182 | One-keystroke capture into a chosen folder and template |+| [Spaced Repetition](https://github.com/st3v3nmw/obsidian-spaced-repetition) | 592,497 | Flashcards from your own notes. See [learning from your vault](../docs/08-outputs/teaching-yourself.md) |+| [Zotero Integration](https://github.com/obsidian-community/obsidian-zotero-integration) | 550,881 | Citations, bibliographies and PDF annotations from Zotero |+| [ReadItLater](https://github.com/DominikPieper/obsidian-ReadItLater) | 134,982 | Saves online content into the vault with a template |+| [Webpage HTML Export](https://github.com/KosmosisDire/obsidian-webpage-export) | 152,194 | Export notes, canvases or the whole vault as HTML |++## A warning about plugin count++Every plugin is code with write access to your notes, running alongside an agent+that also writes to them. Two things that rewrite files on their own schedule+will eventually disagree.++Install what solves a problem you actually have. An empty vault with fifteen+plugins is a procrastination artifact.diff --git a/resources/repositories.md b/resources/repositories.mdindex fcac846..4397ef5 100644--- a/resources/repositories.md+++ b/resources/repositories.md@@ -1,110 +1,65 @@# Repositories-Beyond the implementations of this pattern in [skills.md](skills.md). Every link-here was opened and checked; anything I could not verify is named without a link-rather than guessed at.--## Graph views for Obsidian--The built-in graph view is a viewer with no notion of relationship types. These-replace it when you want the graph to do work.--- **[HEmile/juggl](https://github.com/HEmile/juggl)**. An interactive, stylable,- expandable graph built on Cytoscape.js. Its workspace mode is the real feature:- you build a graph of just the nodes relevant to what you are working on, expand- and hide selectively, pin positions, and save it for later. Supports link types- on edges, which plain Obsidian does not. Has an API other plugins build on.-- **[zsviczian/excalibrain](https://github.com/zsviczian/excalibrain)**. A- structured mind-map of the vault rather than a force-directed cloud. It derives- five relationship types from your links, Dataview fields, tags and frontmatter:- parents, children, friends, other friends, siblings. Requires Dataview and- Excalidraw. The closest thing to [typed- links](../docs/05-graphs/typed-links.md) with no extra work from you.-- **[michaelpporter/breadcrumbs](https://github.com/michaelpporter/breadcrumbs)**.- Typed links plus navigation over them: breadcrumb trails, tree and matrix- views, Mermaid and Markmap rendering, export to Canvas. Reads typed frontmatter- links, tags, lists, folder notes and Dataview queries, and derives implied- relations. Maintained by michaelpporter since May 2026, originally by- SkepticMystic.-- **[SkepticMystic/graph-analysis](https://github.com/SkepticMystic/graph-analysis)**.- Runs actual graph algorithms over your vault, including similarity measures- like Adamic Adar, to surface connections you never linked. The closest thing to- the [metrics](../docs/05-graphs/metrics.md) page without leaving Obsidian.-- **[AlexW00/obsidian-3d-graph](https://github.com/AlexW00/obsidian-3d-graph)**.- The vault as a 3D force graph, built on D3 and 3d-force-graph. Genuinely useful- for spotting cluster structure that a 2D hairball hides, and the source of most- good vault screenshots. The actively maintained fork is published as "3D Graph- New".-- **[brianpetro/obsidian-smart-connections](https://github.com/brianpetro/obsidian-smart-connections)**.- Semantic similarity between notes using a local embedding model, no API key,- shown as a graph and a list while you write. This is the embeddings side of- [graph vs embeddings](../docs/05-graphs/graph-vs-vectors.md) in the one place- it clearly earns its keep: suggesting links you have not made yet.--## Building a graph from your own material--- **[Graphify-Labs/graphify](https://github.com/Graphify-Labs/graphify)**. A- `/graphify` skill for Claude Code, Cursor, Codex and Gemini CLI that reads a- folder and builds a queryable knowledge graph from it. Local deterministic AST- parsing for code, and multimodal extraction for docs, PDFs, screenshots and- diagrams. No vector store, and every edge is labelled as extracted or inferred,- which is the same honesty about provenance this guide asks of page contracts.-- Directly relevant here: it was built around the problem of a `raw/` folder full- of papers, screenshots and notes, which is exactly the folder this pattern- creates. Output includes an interactive `graph.html` and an Obsidian-openable- vault. The project reports large token savings versus reading raw files; that- figure is self-reported and depends heavily on corpus size.-- Originally by Safi Shamsi. Several widely linked repos named `graphify` are- forks of it.--## GraphRAG implementations--Read these if you want to see what an automated version of the graph looks like-before deciding you do not need one. See-[GraphRAG and where it fits](../docs/05-graphs/graphrag.md).--- **[microsoft/graphrag](https://github.com/microsoft/graphrag)**. The reference- implementation of the [paper](papers.md). Entity extraction, community- detection, community summaries, global and local search.-- **[gusye1234/nano-graphrag](https://github.com/gusye1234/nano-graphrag)**. The- same idea in roughly 1,100 lines, written to be read and modified. This is the- one to open if you want to understand the pipeline rather than run it.- Pluggable storage: networkx by default, Neo4j available.-- **[HKUDS/LightRAG](https://github.com/HKUDS/LightRAG)**. A dual-layer approach- holding both a knowledge graph and vector embeddings, aimed at cheaper- incremental updates than the original. Published at EMNLP 2025. Structurally- based on nano-graphrag.-- **[DEEP-PolyU/Awesome-GraphRAG](https://github.com/DEEP-PolyU/Awesome-GraphRAG)**.- A curated list of surveys, papers, benchmarks and open-source projects in this- area. The right starting point if you want the landscape rather than one tool.--Also worth knowing by name: Graphiti and Cognee, both graph-based memory layers-for agents rather than knowledge bases for people. Different problem, adjacent-techniques.--## Publishing--- **[jackyzha0/quartz](https://github.com/jackyzha0/quartz)**. Publishes a vault- as a static site with backlinks, link previews, a local graph and full-text- search. Understands both wikilinks and markdown links, which is the thing to- check in any publishing tool. Docs at- [quartz.jzhao.xyz](https://quartz.jzhao.xyz/).-- One caveat that applies to every publishing tool: check how it handles links to- unpublished pages before you publish anything. See [publishing and- export](../docs/08-outputs/publishing-and-export.md).+Graph, retrieval and memory projects worth knowing. Stars from the GitHub API,+September 2026. For agent skills see [skills.md](skills.md); for Obsidian+plugins see [plugins.md](plugins.md).-## How to judge one of these+## GraphRAG and graph retrieval++| Repo | Stars | What it is |+|---|---|---|+| [HKUDS/LightRAG](https://github.com/HKUDS/LightRAG) | 39,467 | Dual-layer graph plus vectors, aimed at cheap incremental updates. EMNLP 2025 |+| [microsoft/graphrag](https://github.com/microsoft/graphrag) | 35,875 | The reference implementation of the [paper](papers.md) |+| [gusye1234/nano-graphrag](https://github.com/gusye1234/nano-graphrag) | 3,983 | The same idea in about 1,100 readable lines. Open this one to understand the pipeline |+| [OSU-NLP-Group/HippoRAG](https://github.com/OSU-NLP-Group/HippoRAG) | 3,985 | The NeurIPS'24 paper's implementation. Personalised PageRank over a knowledge graph |+| [DEEP-PolyU/Awesome-GraphRAG](https://github.com/DEEP-PolyU/Awesome-GraphRAG) | 2,635 | Curated map of surveys, benchmarks and projects in the area |++## Agent memory++Adjacent problem, same techniques. These give an agent persistent memory; this+guide gives a person a knowledge base. Worth reading for the retrieval design+even if you never run one.++| Repo | Stars | What it is |+|---|---|---|+| [mem0ai/mem0](https://github.com/mem0ai/mem0) | 64,863 | Drop-in memory layer for agents and apps |+| [getzep/graphiti](https://github.com/getzep/graphiti) | 30,680 | Real-time temporal knowledge graphs for agents |+| [topoteretes/cognee](https://github.com/topoteretes/cognee) | 30,568 | Turns data into knowledge graphs, combining graph and vector retrieval |+| [doobidoo/mcp-memory-service](https://github.com/doobidoo/mcp-memory-service) | 1,928 | Persistent memory over MCP, REST API plus knowledge graph |++## RAG frameworks++If you decide your vault has outgrown files. See [adding RAG when you need+it](../docs/07-retrieval/rag-on-top.md) for the threshold, which most personal+vaults never reach.-Two questions answer most of it.+| Repo | Stars | What it is |+|---|---|---|+| [infiniflow/ragflow](https://github.com/infiniflow/ragflow) | 90,216 | Full RAG engine with document understanding |+| [run-llama/llama_index](https://github.com/run-llama/llama_index) | 52,058 | Document agent and ingestion framework |+| [neuml/txtai](https://github.com/neuml/txtai) | 12,931 | Embeddings database and LLM workflows, small enough to read |++## MCP++| Repo | Stars | What it is |+|---|---|---|+| [punkpeye/awesome-mcp-servers](https://github.com/punkpeye/awesome-mcp-servers) | 94,575 | The index of MCP servers |+| [modelcontextprotocol/servers](https://github.com/modelcontextprotocol/servers) | 90,141 | The official reference servers, including filesystem and memory |+| [MarkusPfundstein/mcp-obsidian](https://github.com/MarkusPfundstein/mcp-obsidian) | 4,376 | Third-party MCP server for Obsidian over the REST plugin |++Since the Local REST API plugin now ships its own MCP server, a separate one is+optional. See [MCP for Obsidian](../docs/02-setup/mcp-obsidian.md).++## How to judge one of these**Does it own your data or read it?** Anything that reads your markdown andwrites derived output is safe to try and safe to abandon. Anything that becomesthe only place a piece of knowledge lives has taken the portability you chosemarkdown for.-**Is it maintained?** Check the commit history, not the star count. This-ecosystem produces a lot of repos in the weeks after a popular gist, and most of-them stop within a month.+**Is it maintained?** Check the last push date, not the star count. This+ecosystem produces a lot of repos in the weeks after a popular gist, and most+stop within a month.++**Are the stars measuring the right thing?** For plugins, installs are the+better signal, and the two diverge by an order of magnitude in places.diff --git a/resources/skills.md b/resources/skills.mdindex 0b7e4ab..22af4c5 100644--- a/resources/skills.md+++ b/resources/skills.md@@ -1,52 +1,67 @@-# Skills and implementations--Agent skills that run this pattern, and open-source builds worth reading before-writing your own.--## In this repo--Four skills in [`skills/`](../skills/README.md): ingest, lint, query, review.-Five commands in [`commands/`](../commands/README.md). Four subagent definitions-in [`agents/`](../agents/README.md), two of them read-only by design.--All plain `SKILL.md` files, so they work with any agent that reads the Agent-Skills format, not only Claude Code.--## Community implementations--Reading someone else's page contracts is the fastest way to improve your own.-These are worth opening even if you never install them.--- **[Astro-Han/karpathy-llm-wiki](https://github.com/Astro-Han/karpathy-llm-wiki)**.- The pattern packaged as a single installable skill for Claude Code, Cursor and- Codex. Ingest into `raw/`, compile into `wiki/`, cited answers, lint. Includes- a design spec and a list of what the author deliberately did not build, which- is the most useful section in it.-- **[micuintus/llm-wiki](https://github.com/micuintus/llm-wiki)**. A deliberately- minimal skill: no dependencies, no JSON metadata, convention-based guardrails- rather than code-based ones. Good counterpoint to the heavier- implementations, and its README lists several others in the same family.-- **[NicholasSpisak/second-brain](https://github.com/NicholasSpisak/second-brain)**.- Four skills installable through npm, close to the original gist, with a setup- wizard.-- **[eugeniughelbur/obsidian-second-brain](https://github.com/eugeniughelbur/obsidian-second-brain)**.- Around 43 ready commands, works across Claude, Codex and Gemini. The author's- write-up of what the gist leaves undefined is worth reading alongside it.--Two more get recommended often and are worth a look if the above do not fit:-`AgriciDaniel/claude-obsidian`, which ships role presets, and-`coleam00/second-brain-starter`, which interviews you and generates a plan-before building anything.+# Skills and agents-## Writing your own+Counted individually: a repo shipping sixteen skills counts as sixteen, because+that is what you install. Counts were taken by reading each repository's file+tree in September 2026.++## This repo++| Type | Count | Where |+|---|---|---|+| Skills | 4 | [`skills/`](../skills/README.md). Ingest, lint, query, review |+| Commands | 5 | [`commands/`](../commands/README.md). `/ingest`, `/ask`, `/lint`, `/review`, `/backfill` |+| Subagents | 4 | [`agents/`](../agents/README.md). Ingestor, linker, reviewer, researcher |+| Scripts | 4 | [`scripts/`](../scripts/README.md). Link check, stats, graph export, chat import |++Plain `SKILL.md` files, so they work with any agent that reads the Agent Skills+format.++## Implementations of this pattern++| Repo | Stars | Ships |+|---|---|---|+| [AgriciDaniel/claude-obsidian](https://github.com/AgriciDaniel/claude-obsidian) | 14,706 | 16 skills, 3 subagents. Self-organizing vault with role presets |+| [Astro-Han/karpathy-llm-wiki](https://github.com/Astro-Han/karpathy-llm-wiki) | 2,174 | 1 skill covering the full ingest, compile, query, lint loop |+| [ballred/obsidian-claude-pkm](https://github.com/ballred/obsidian-claude-pkm) | 1,856 | 13 skills, 4 subagents. A complete starter kit |+| [coleam00/second-brain-starter](https://github.com/coleam00/second-brain-starter) | 768 | 1 skill that interviews you and generates a build plan |+| [NicholasSpisak/second-brain](https://github.com/NicholasSpisak/second-brain) | 704 | 4 skills, npm installer, close to the original gist |+| [eugeniughelbur/obsidian-second-brain](https://github.com/eugeniughelbur/obsidian-second-brain) | small | 47 commands across Claude, Codex and Gemini |+| [micuintus/llm-wiki](https://github.com/micuintus/llm-wiki) | small | 1 skill, deliberately minimal, no dependencies. Good counterpoint |++The two at the bottom are small on purpose and still worth reading. Star count+measures reach, not quality, and in this corner of the ecosystem it mostly+measures who posted about it.-The format is a `SKILL.md` with frontmatter and a body. The four in this repo-follow a consistent shape: why the skill exists, one core rule, a workflow, an-output template, calibration, and a worked example.+## General skill libraries++Not second-brain specific, but this is where the format itself is defined and+where the best-written examples live.++| Repo | Stars | Ships |+|---|---|---|+| [obra/superpowers](https://github.com/obra/superpowers) | 282,827 | 14 skills. An agentic skills framework and development methodology |+| [anthropics/skills](https://github.com/anthropics/skills) | 175,035 | 20 skills, 3 subagents. The official reference for the format |+| [VoltAgent/awesome-agent-skills](https://github.com/VoltAgent/awesome-agent-skills) | 33,894 | A curated index of 1,000+ community skills |+| [hesreallyhim/awesome-claude-code](https://github.com/hesreallyhim/awesome-claude-code) | 53,661 | Commands, hooks, workflows and tooling for Claude Code |++## Graph builders++| Repo | Stars | What it does |+|---|---|---|+| [Graphify-Labs/graphify](https://github.com/Graphify-Labs/graphify) | 115,697 | A `/graphify` skill that turns any folder of code, docs, PDFs and screenshots into a queryable graph. Every edge labelled extracted or inferred |++Graphify is the one to try first if you already have a full `raw/` folder. It+was built around exactly that problem, it runs locally with deterministic+parsing for code, and it outputs an interactive `graph.html` plus an+Obsidian-openable vault. Its reported token savings are self-reported and depend+heavily on corpus size.++## Writing your own-Two authoring rules that matter more than the rest. Explain the reason behind-every rule, because rules cover the cases you thought of and reasons cover the-rest. And name the near-misses where the skill should not fire, because-over-triggering is what makes people uninstall skills.+Three properties make a skill worth writing: you do it repeatedly, you have+opinions about how, and the opinions are not obvious enough for a model to+guess.-Details in [skills and commands](../docs/06-agents/skills-and-commands.md).+Read `anthropics/skills` for the format and one of the small implementations+above for a full worked example. Details in [skills and+commands](../docs/06-agents/skills-and-commands.md).diff --git a/resources/tools.md b/resources/tools.mdindex 0187f04..bead93f 100644--- a/resources/tools.md+++ b/resources/tools.md@@ -1,74 +1,70 @@# Tools+Everything outside the Obsidian plugin catalog. For plugins, see+[plugins.md](plugins.md). Star counts are from GitHub, September 2026.+## The core stack-- **[Obsidian](https://obsidian.md)**. Local markdown editor. The vault is a- plain folder, which is the whole reason agents work well with it.-- **[Obsidian Web Clipper](https://obsidian.md/clipper)**. Official browser- extension for saving articles into the vault, built by the Obsidian team. Uses- Mozilla Readability, the engine behind Firefox Reader View.-- **[Claude Code](https://code.claude.com/docs/en/setup)**. The agent that- maintains the wiki. Runs in the vault folder and edits files directly.- Requires a paid account; the free plan does not include access.+| Tool | What it does |+|---|---|+| [Obsidian](https://obsidian.md) | Local markdown editor. The vault is a plain folder, which is why agents work well with it |+| [Obsidian Web Clipper](https://obsidian.md/clipper) | Official browser extension, saves articles into `raw/`. Uses Mozilla Readability |+| [Claude Code](https://code.claude.com/docs/en/setup) | The agent that maintains the wiki. Paid account required |++## Alternative homes for a vault-## Obsidian plugins+| Tool | Stars | Notes |+|---|---|---|+| [AFFiNE](https://github.com/toeverything/AFFiNE) | 72,292 | Docs, whiteboard and database in one. Heavier than markdown files |+| [memos](https://github.com/usememos/memos) | 62,837 | Self-hosted quick capture, markdown-native. A good `raw/` inbox |+| [Joplin](https://github.com/laurent22/joplin) | 56,264 | Privacy-focused notes with sync across every platform |+| [SiYuan](https://github.com/siyuan-note/siyuan) | 46,225 | Self-hosted knowledge workspace, block-based |+| [Logseq](https://github.com/logseq/logseq) | 44,818 | Outliner, plain markdown, block-level references |+| [Outline](https://github.com/outline/outline) | 40,483 | Team knowledge base. The right answer when it is not just you |+| [Trilium](https://github.com/TriliumNext/Trilium) | 37,762 | Hierarchical personal knowledge base with scripting |+| [Foam](https://github.com/foambubble/foam) | 17,386 | The same wikilink pattern inside VS Code |+| [Anytype](https://github.com/anyproto/anytype-ts) | 8,764 | Local-first, encrypted, object-based |+| [Dendron](https://github.com/dendronhq/dendron) | 7,465 | Hierarchical PKM in VS Code |-- **[Dataview](https://github.com/blacksmithgu/obsidian-dataview)**. A query- language over your frontmatter, rendered inline. This is what answers- structural questions like "which concept pages have not been touched in ninety- days". Docs at- [blacksmithgu.github.io/obsidian-dataview](https://blacksmithgu.github.io/obsidian-dataview/).-- **[Templater](https://github.com/SilentVoid13/Templater)**. Templates with- variables and dates, for the page shapes in `vault-template/templates/`. The- agent does not need it; you will, for pages you write by hand.-- **[Local REST API](https://github.com/coddingtonbear/obsidian-local-rest-api)**.- Serves your vault over a local API. Worth reading the README before choosing- an MCP route: the plugin now ships a **built-in MCP server** at `/mcp/`, which- removes the need for a separate server in most setups.+Worth being honest about the tradeoff: everything below Obsidian in this table+stores knowledge in its own structure. Markdown files in a folder are the only+format where leaving the tool costs nothing.-## MCP+## AI-native knowledge apps-- **[mcp-obsidian](https://github.com/MarkusPfundstein/mcp-obsidian)**. The- widely used third-party MCP server, run via `uvx mcp-obsidian`, talking to the- Local REST API plugin. Still the most documented path, and the one most guides- including this one describe.+| Tool | Stars | Notes |+|---|---|---|+| [Khoj](https://github.com/khoj-ai/khoj) | 37,186 | Self-hostable AI second brain over your docs, with custom agents and scheduled automations |+| [Quivr](https://github.com/The-Vibe-Company/quivr) | 39,498 | Opinionated RAG you embed in your own app |+| [Reor](https://github.com/reorproject/reor) | 8,563 | Local-first AI note app that links notes automatically as you write |-Worth knowing before you pick: at least one maintained fork-([proofsh/obsidian-mcp](https://github.com/proofsh/obsidian-mcp)) has been-archived with the explicit reasoning that routing through a REST plugin adds-complexity when the vault is markdown files on disk. That is the same argument-this guide makes for starting on the filesystem.+These are products, not patterns. They do the job for you, at the cost of the+portability this guide is built around. Worth knowing before you decide the+file-based approach is too much work.-## Capture+## Capture and processing-- **[yt-dlp](https://github.com/yt-dlp/yt-dlp)**. Subtitles and audio from video- platforms. `--write-auto-sub --skip-download` is the flag pair you want.-- **[youtube-transcript-api](https://github.com/jdepoix/youtube-transcript-api)**.- Python library for pulling transcripts directly.-- **[OCRmyPDF](https://github.com/ocrmypdf/OCRmyPDF)**. Adds a text layer to- scanned PDFs without changing the visible document.-- **[Zotero](https://www.zotero.org)**. Reference manager with a markdown export- path, worth it if you read papers regularly.-- **[Readwise](https://readwise.io)**. Kindle and read-later highlights, syncs- into Obsidian. Paid.+| Tool | What it does |+|---|---|+| [yt-dlp](https://github.com/yt-dlp/yt-dlp) | Subtitles and audio from video platforms |+| [youtube-transcript-api](https://github.com/jdepoix/youtube-transcript-api) | Transcripts from Python |+| [OCRmyPDF](https://github.com/ocrmypdf/OCRmyPDF) | Text layer for scanned PDFs |+| [Zotero](https://github.com/zotero/zotero) | Reference manager, 15,184 stars, markdown export path |+| [Readwise](https://readwise.io) | Kindle and read-later highlights into Obsidian. Paid |## Search and graph-- **[ripgrep](https://github.com/BurntSushi/ripgrep)**. Fast enough that most- vaults never need an index.-- **[NetworkX](https://networkx.org)**. Graph analysis in Python once you have- exported an edge list.-- **[Kuzu](https://kuzudb.com)**. Embedded graph database, no server to run.-- **[Neo4j](https://neo4j.com)**. The full version. Overkill unless you are- building something on top.-- **[Gephi](https://gephi.org)**. Visual graph exploration beyond what Obsidian's- graph view does.+| Tool | What it does |+|---|---|+| [ripgrep](https://github.com/BurntSushi/ripgrep) | Fast enough that most vaults never need an index |+| [NetworkX](https://networkx.org) | Graph analysis once you have exported an edge list |+| [Kuzu](https://kuzudb.com) | Embedded graph database, no server |+| [Neo4j](https://neo4j.com) | Full graph database. Overkill unless you build on top |+| [Gephi](https://gephi.org) | Visual graph exploration beyond Obsidian's view |+| [txtai](https://github.com/neuml/txtai) | 12,931 stars. Semantic search and LLM workflows, if you decide you need embeddings |-## Alternatives to Obsidian+## Publishing-- **[Logseq](https://logseq.com)**. Outliner, also plain markdown, block-level- references.-- **[Foam](https://foambubble.github.io/foam/)**. The same pattern inside VS- Code.-- **Plain folders plus git.** Genuinely viable. Obsidian is a viewer for the- graph, not a requirement of it.+| Tool | Stars | Notes |+|---|---|---|+| [Quartz](https://github.com/jackyzha0/quartz) | 13,185 | Vault to static site with backlinks, local graph and search |