initial flake app

- glob.ts: add optional respectGitignore param (default true) so callers
  can opt out of gitignore filtering for generated directories
- build.ts: pass respectGitignore=false when globbing content/ so
  gitignored-but-generated markdown files are always picked up
- static.ts: copy top-level static/ → output/ root with gitignore disabled,
  mirroring Hugo's convention so ox-hugo images at static/ox-hugo/* are
  served at /ox-hugo/* as expected by the exported markdown src paths
- oxhugofm.ts: run replaceFigureWithMdImg before removeHugoShortcode so the
  precise figure regex sees the intact shortcode before the generic {{.*}}
  stripper destroys it; also upgrade figureTagRegex to figureShortcodeRegex
  which matches the full shortcode form ox-hugo actually emits

.gitignore

@@ -9,3 +9,16 @@ tsconfig.tsbuildinfo
 private/
 .replit
 replit.nix
+erl_crash.dump
+# content/ is generated by the export script; only keep the placeholder
+content/*
+!content/.gitkeep
+# static/ox-hugo/ is populated by ox-hugo during export
+static/ox-hugo/
+# Elixir/Mix build artifacts for the pipeline project
+scripts/pipeline/_build/
+scripts/pipeline/deps/
+scripts/pipeline/erl_crash.dump
+# Test helpers (not needed in production)
+scripts/test.bib
+scripts/test_pipeline.exs

AGENTS.md

@@ -0,0 +1,361 @@
+# AGENTS.md - Coding Agent Instructions
+
+This document provides essential information for AI coding agents working in this repository.
+
+## Project Overview
+
+**Quartz** is a static site generator for publishing digital gardens and notes as websites.
+Built with TypeScript, Preact, and unified/remark/rehype for markdown processing.
+
+| Stack         | Technology                                |
+| ------------- | ----------------------------------------- |
+| Language      | TypeScript 5.x (strict mode)              |
+| Runtime       | Node.js >=22 (v22.16.0 pinned)            |
+| Package Mgr   | npm >=10.9.2                              |
+| Module System | ES Modules (`"type": "module"`)           |
+| UI Framework  | Preact 10.x (JSX with `react-jsx` pragma) |
+| Build Tool    | esbuild                                   |
+| Styling       | SCSS via esbuild-sass-plugin              |
+
+## Environment
+
+This is a Nix project. Use the provided `flake.nix` to enter a dev shell with Node.js 22 and npm:
+
+```bash
+nix develop
+```
+
+All `npm` commands below must be run inside the dev shell.
+
+## Build, Lint, and Test Commands
+
+```bash
+# Type check and format check (CI validation)
+npm run check
+
+# Auto-format code with Prettier
+npm run format
+
+# Run all tests
+npm run test
+
+# Run a single test file
+npx tsx --test quartz/util/path.test.ts
+
+# Run tests matching a pattern (use --test-name-pattern)
+npx tsx --test --test-name-pattern="typeguards" quartz/util/path.test.ts
+
+# Build the static site
+npx quartz build
+
+# Build and serve with hot reload
+npx quartz build --serve
+
+# Profile build performance
+npm run profile
+```
+
+### Test Files Location
+
+Tests use Node.js native test runner via `tsx`. Test files follow the `*.test.ts` pattern:
+
+- `quartz/util/path.test.ts`
+- `quartz/util/fileTrie.test.ts`
+- `quartz/components/scripts/search.test.ts`
+
+## Code Style Guidelines
+
+### Prettier Configuration (`.prettierrc`)
+
+```json
+{
+  "printWidth": 100,
+  "tabWidth": 2,
+  "semi": false,
+  "trailingComma": "all",
+  "quoteProps": "as-needed"
+}
+```
+
+**No ESLint** - only Prettier for formatting. Run `npm run format` before committing.
+
+### TypeScript Configuration
+
+- **Strict mode enabled** (`strict: true`)
+- `noUnusedLocals: true` - no unused variables
+- `noUnusedParameters: true` - no unused function parameters
+- JSX configured for Preact (`jsxImportSource: "preact"`)
+
+### Import Conventions
+
+```typescript
+// 1. External packages first
+import { PluggableList } from "unified"
+import { visit } from "unist-util-visit"
+
+// 2. Internal utilities/types (relative paths)
+import { QuartzTransformerPlugin } from "../types"
+import { FilePath, slugifyFilePath } from "../../util/path"
+import { i18n } from "../../i18n"
+```
+
+### Naming Conventions
+
+| Element          | Convention   | Example                             |
+| ---------------- | ------------ | ----------------------------------- |
+| Files (utils)    | camelCase    | `path.ts`, `fileTrie.ts`            |
+| Files (comps)    | PascalCase   | `TableOfContents.tsx`, `Search.tsx` |
+| Types/Interfaces | PascalCase   | `QuartzComponent`, `FullSlug`       |
+| Type Guards      | `is*` prefix | `isFilePath()`, `isFullSlug()`      |
+| Constants        | UPPER_CASE   | `QUARTZ`, `UPSTREAM_NAME`           |
+| Options types    | `Options`    | `interface Options { ... }`         |
+
+### Branded Types Pattern
+
+This codebase uses branded types for type-safe path handling:
+
+```typescript
+type SlugLike<T> = string & { __brand: T }
+export type FilePath = SlugLike<"filepath">
+export type FullSlug = SlugLike<"full">
+export type SimpleSlug = SlugLike<"simple">
+
+// Always validate with type guards before using
+export function isFilePath(s: string): s is FilePath { ... }
+```
+
+### Component Pattern (Preact)
+
+Components use a factory function pattern with attached static properties:
+
+```typescript
+export default ((userOpts?: Partial<Options>) => {
+  const opts: Options = { ...defaultOptions, ...userOpts }
+
+  const ComponentName: QuartzComponent = ({ cfg, displayClass }: QuartzComponentProps) => {
+    return <div class={classNames(displayClass, "component-name")}>...</div>
+  }
+
+  ComponentName.css = style           // SCSS styles
+  ComponentName.afterDOMLoaded = script  // Client-side JS
+  return ComponentName
+}) satisfies QuartzComponentConstructor
+```
+
+### Plugin Pattern
+
+Three plugin types: transformers, filters, and emitters.
+
+```typescript
+export const PluginName: QuartzTransformerPlugin<Partial<Options>> = (userOpts) => {
+  const opts = { ...defaultOptions, ...userOpts }
+  return {
+    name: "PluginName",
+    markdownPlugins(ctx) { return [...] },
+    htmlPlugins(ctx) { return [...] },
+    externalResources(ctx) { return { js: [], css: [] } },
+  }
+}
+```
+
+### Testing Pattern
+
+Use Node.js native test runner with `assert`:
+
+```typescript
+import test, { describe, beforeEach } from "node:test"
+import assert from "node:assert"
+
+describe("FeatureName", () => {
+  test("should do something", () => {
+    assert.strictEqual(actual, expected)
+    assert.deepStrictEqual(actualObj, expectedObj)
+    assert(condition) // truthy assertion
+    assert(!condition) // falsy assertion
+  })
+})
+```
+
+### Error Handling
+
+- Use `try/catch` for critical operations (file I/O, parsing)
+- Custom `trace` utility for error reporting with stack traces
+- `process.exit(1)` for fatal errors
+- `console.warn()` for non-fatal issues
+
+### Async Patterns
+
+- Prefer `async/await` over raw promises
+- Use async generators (`async *emit()`) for streaming file output
+- Use `async-mutex` for concurrent build protection
+
+## Project Structure
+
+```
+quartz/
+├── bootstrap-cli.mjs    # CLI entry point
+├── build.ts             # Build orchestration
+├── cfg.ts               # Configuration types
+├── components/          # Preact UI components
+│   ├── *.tsx            # Components
+│   ├── scripts/         # Client-side scripts (*.inline.ts)
+│   └── styles/          # Component SCSS
+├── plugins/
+│   ├── transformers/    # Markdown AST transformers
+│   ├── filters/         # Content filters
+│   ├── emitters/        # Output generators
+│   └── types.ts         # Plugin type definitions
+├── processors/          # Build pipeline (parse/filter/emit)
+├── util/                # Utility functions
+└── i18n/                # Internationalization (30+ locales)
+```
+
+## Branch Workflow
+
+This is a fork of [jackyzha0/quartz](https://github.com/jackyzha0/quartz) with org-roam customizations.
+
+| Branch      | Purpose                                          |
+| ----------- | ------------------------------------------------ |
+| `main`      | Clean mirror of upstream quartz — no custom code |
+| `org-roam`  | Default branch — all customizations live here    |
+| `feature/*` | Short-lived branches off `org-roam`              |
+
+### Pulling Upstream Updates
+
+```bash
+git checkout main
+git fetch upstream
+git merge upstream/main
+git checkout org-roam
+git merge main
+# Resolve conflicts if any, then commit
+```
+
+### Working on Features
+
+```bash
+git checkout org-roam
+git checkout -b feature/my-feature
+# ... work ...
+git checkout org-roam
+git merge feature/my-feature
+git branch -d feature/my-feature
+```
+
+**Merge direction:** `upstream → main → org-roam → feature/*`
+
+## Org-Roam Workflow
+
+Notes live in a **separate directory** outside this repo. The export pipeline
+converts them to Markdown via ox-hugo, applies post-processing transforms, then
+Quartz builds the site.
+
+### Tooling
+
+The dev shell (`nix develop`) provides:
+
+- `nodejs_22` — Quartz build
+- `elixir` — runs the export script and pipeline
+- `emacs` + `ox-hugo` — performs the org → markdown conversion
+
+### Export and build
+
+```bash
+# Export only (wipes content/, exports all .org files, runs pipeline)
+NOTES_DIR=/path/to/notes npm run export
+
+# Export then build the site
+NOTES_DIR=/path/to/notes npm run build:notes
+
+# Positional arg also works
+elixir scripts/export.exs /path/to/notes
+```
+
+Optional env vars for the pipeline:
+
+| Var             | Default                  | Purpose                                   |
+| --------------- | ------------------------ | ----------------------------------------- |
+| `BIBTEX_FILE`   | —                        | Path to `.bib` file for citation fallback |
+| `ZOTERO_URL`    | `http://localhost:23119` | Zotero Better BibTeX base URL             |
+| `CITATION_MODE` | `warn`                   | `silent` / `warn` / `strict`              |
+
+### Export pipeline phases
+
+`scripts/export.exs` runs four phases in sequence:
+
+1. **Wipe** `content/` (preserving `.gitkeep`)
+2. **Export** each `.org` file via `emacs --batch` + `ox-hugo` → `content/**/*.md`
+3. **Pipeline** — run Elixir transform modules over every `.md` file
+4. **Index** — generate a fallback `content/index.md` if none was exported
+
+The export uses TOML frontmatter (`+++`) and per-file mode (not per-subtree).
+
+### Markdown pipeline (`scripts/pipeline/`)
+
+A standalone Mix project that post-processes `content/*.md` after ox-hugo.
+It is compiled automatically on first run; subsequent runs use the `_build/`
+cache and are fast.
+
+**Architecture:**
+
+```
+scripts/pipeline/
+├── mix.exs                          # deps: req, jason
+└── lib/
+    ├── pipeline.ex                  # Generic runner (fold transforms over .md files)
+    ├── pipeline/
+    │   ├── application.ex           # OTP app — starts Finch HTTP pool
+    │   ├── transform.ex             # Behaviour: init/1, apply/3, teardown/1
+    │   ├── transforms/
+    │   │   └── citations.ex         # Resolves cite:key → [Label](url)
+    │   └── resolvers/
+    │       ├── zotero.ex            # JSON-RPC to Zotero Better BibTeX
+    │       ├── bibtex.ex            # Parses local .bib file
+    │       └── doi.ex               # Bare-key fallback (always succeeds)
+```
+
+**Adding a new transform:**
+
+1. Create `scripts/pipeline/lib/pipeline/transforms/my_transform.ex`
+2. Implement the `Pipeline.Transform` behaviour (`init/1`, `apply/3`)
+3. Append the module to `transforms` in `scripts/export.exs`
+
+```elixir
+transforms = [
+  Pipeline.Transforms.Citations,
+  Pipeline.Transforms.MyTransform,  # new
+]
+```
+
+### Citation resolution (`Pipeline.Transforms.Citations`)
+
+Handles org-citar syntax that passes through ox-hugo unchanged:
+
+| Syntax           | Example              |
+| ---------------- | -------------------- |
+| org-cite / citar | `[cite:@key]`        |
+| multiple keys    | `[cite:@key1;@key2]` |
+| bare (legacy)    | `cite:key`           |
+
+Resolution chain (first success wins):
+
+1. **Zotero** — JSON-RPC to `localhost:23119/better-bibtex/json-rpc`
+   - Calls `item.search` to find the item, then `item.attachments` to get
+     the PDF link (`zotero://open-pdf/library/items/KEY`)
+   - Falls back to `zotero://select/library/items/KEY` if no PDF attachment
+   - Probe uses a JSON-RPC call, **not** `/better-bibtex/cayw`
+     (that endpoint blocks waiting for interactive input)
+2. **BibTeX** — parses `BIBTEX_FILE`; extracts authors, year, DOI/URL
+3. **DOI fallback** — always succeeds; renders bare key or `https://doi.org/...`
+
+**Zotero JSON-RPC gotcha:** `Req 0.5` does not allow combining `:finch` and
+`:connect_options` in the same call. Use `:receive_timeout` only.
+
+## Important Notes
+
+- **Client-side scripts**: Use `.inline.ts` suffix, bundled via esbuild
+- **Isomorphic code**: `quartz/util/path.ts` must not use Node.js APIs
+- **Incremental builds**: Plugins can implement `partialEmit` for efficiency
+- **Markdown flavors**: Supports Obsidian (`ofm.ts`) and Roam (`roam.ts`) syntax
+- **Pipeline build artifacts**: `scripts/pipeline/_build/` and `scripts/pipeline/deps/`
+  are gitignored — run `mix deps.get` inside `scripts/pipeline/` after a fresh clone

README.md

@@ -1,13 +1,96 @@
-# Quartz v4
+# Quartz v4 — org-roam edition
 
-> “[One] who works with the door open gets all kinds of interruptions, but [they] also occasionally gets clues as to what the world is and what might be important.” — Richard Hamming
+> "[One] who works with the door open gets all kinds of interruptions, but [they] also occasionally gets clues as to what the world is and what might be important." — Richard Hamming
 
 Quartz is a set of tools that helps you publish your [digital garden](https://jzhao.xyz/posts/networked-thought) and notes as a website for free.
 
-🔗 Read the documentation and get started: https://quartz.jzhao.xyz/
+This fork adds first-class support for [org-roam](https://www.orgroam.com/) notes via [ox-hugo](https://ox-hugo.scripter.co/).
+
+🔗 Upstream documentation: https://quartz.jzhao.xyz/
 
 [Join the Discord Community](https://discord.gg/cRFFHYye7t)
 
+## Quick Start
+
+### Prerequisites
+
+This project uses Nix. Enter the development shell, which provides Node.js 22, Elixir, and Emacs with ox-hugo:
+
+```bash
+nix develop
+```
+
+All commands below must be run inside this shell.
+
+```bash
+npm install
+```
+
+### Building from org-roam notes
+
+Your org-roam notes live in a separate directory. Point `NOTES_DIR` at it:
+
+```bash
+# Export notes to content/ and build the site
+NOTES_DIR=/path/to/notes npm run build:notes
+
+# Export, build, and serve with hot reload
+NOTES_DIR=/path/to/notes npm run serve:notes
+
+# Export only (wipes content/ and re-exports all .org files)
+NOTES_DIR=/path/to/notes npm run export
+```
+
+The export pipeline runs in four phases:
+
+1. **Wipe** `content/` clean
+2. **Export** every `.org` file via `emacs --batch` + ox-hugo → Markdown
+3. **Transform** — post-process the Markdown (citation resolution, etc.)
+4. **Index** — generate a fallback `index.md` if none was exported
+
+#### Citations (org-citar → Zotero links)
+
+org-citar references (`[cite:@key]`) are resolved to clickable Zotero links.
+With Zotero running and the [Better BibTeX](https://retorque.re/zotero-better-bibtex/)
+plugin installed, no extra configuration is needed — the pipeline detects it
+automatically and links directly to the PDF in your library.
+
+```bash
+# Use a local .bib file as fallback when Zotero is not running
+BIBTEX_FILE=/path/to/refs.bib NOTES_DIR=/path/to/notes npm run export
+
+# Control warning verbosity for unresolved keys
+CITATION_MODE=strict NOTES_DIR=/path/to/notes npm run export
+```
+
+| Env var         | Default                  | Purpose                                   |
+| --------------- | ------------------------ | ----------------------------------------- |
+| `BIBTEX_FILE`   | —                        | Path to `.bib` file for citation fallback |
+| `ZOTERO_URL`    | `http://localhost:23119` | Zotero Better BibTeX base URL             |
+| `CITATION_MODE` | `warn`                   | `silent` / `warn` / `strict`              |
+
+### Building without org-roam notes
+
+If you manage `content/` directly with Markdown files:
+
+```bash
+# Build the site
+npx quartz build
+
+# Build and serve with hot reload
+npx quartz build --serve
+```
+
+The site is generated in `public/`. When serving, visit http://localhost:8080.
+
+### Development
+
+```bash
+npm run check    # type check + format check
+npm run format   # auto-format with Prettier
+npm run test     # run tests
+```
+
 ## Sponsors
 
 <p align="center">

flake.lock

@@ -0,0 +1,61 @@
+{
+  "nodes": {
+    "flake-utils": {
+      "inputs": {
+        "systems": "systems"
+      },
+      "locked": {
+        "lastModified": 1731533236,
+        "narHash": "sha256-l0KFg5HjrsfsO/JpG+r7fRrqm12kzFHyUHqHCVpMMbI=",
+        "owner": "numtide",
+        "repo": "flake-utils",
+        "rev": "11707dc2f618dd54ca8739b309ec4fc024de578b",
+        "type": "github"
+      },
+      "original": {
+        "owner": "numtide",
+        "repo": "flake-utils",
+        "type": "github"
+      }
+    },
+    "nixpkgs": {
+      "locked": {
+        "lastModified": 1771008912,
+        "narHash": "sha256-gf2AmWVTs8lEq7z/3ZAsgnZDhWIckkb+ZnAo5RzSxJg=",
+        "owner": "NixOS",
+        "repo": "nixpkgs",
+        "rev": "a82ccc39b39b621151d6732718e3e250109076fa",
+        "type": "github"
+      },
+      "original": {
+        "owner": "NixOS",
+        "ref": "nixos-unstable",
+        "repo": "nixpkgs",
+        "type": "github"
+      }
+    },
+    "root": {
+      "inputs": {
+        "flake-utils": "flake-utils",
+        "nixpkgs": "nixpkgs"
+      }
+    },
+    "systems": {
+      "locked": {
+        "lastModified": 1681028828,
+        "narHash": "sha256-Vy1rq5AaRuLzOxct8nz4T6wlgyUR7zLU309k9mBC768=",
+        "owner": "nix-systems",
+        "repo": "default",
+        "rev": "da67096a3b9bf56a91d16901293e51ba5b49a27e",
+        "type": "github"
+      },
+      "original": {
+        "owner": "nix-systems",
+        "repo": "default",
+        "type": "github"
+      }
+    }
+  },
+  "root": "root",
+  "version": 7
+}

flake.nix

@@ -0,0 +1,99 @@
+{
+  description = "Quartz org-roam dev shell and build app";
+
+  inputs = {
+    nixpkgs.url = "github:NixOS/nixpkgs/nixos-unstable";
+    flake-utils.url = "github:numtide/flake-utils";
+  };
+
+  outputs = { self, nixpkgs, flake-utils }:
+    flake-utils.lib.eachDefaultSystem (system:
+      let
+        pkgs = import nixpkgs { inherit system; };
+
+        # Emacs with ox-hugo — shared between devShell and buildApp
+        emacsWithOxHugo = (pkgs.emacsPackagesFor pkgs.emacs-nox).emacsWithPackages
+          (epkgs: [ epkgs.ox-hugo ]);
+
+        # Pre-fetched npm dependency tree (node_modules)
+        quartzDeps = pkgs.buildNpmPackage {
+          pname = "quartz-deps";
+          version = "4.5.2";
+          src = ./.;
+          npmDepsHash = "sha256-7u+VlIx44B3/ivM9vLMIOn+e4TL4eS6B682vhS+Ikb4=";
+          dontBuild = true;
+          installPhase = ''
+            mkdir -p $out
+            cp -r node_modules $out/node_modules
+          '';
+        };
+
+        # Pre-fetched Hex/Mix dependencies for scripts/pipeline
+        pipelineMixDeps = pkgs.beamPackages.fetchMixDeps {
+          pname = "pipeline-mix-deps";
+          version = "0.1.0";
+          src = ./scripts/pipeline;
+          sha256 = "sha256-E79X+nUy86G1Jrwv3T7dXekoGv8Hd14ZgJSKWjvlmAw=";
+        };
+
+        # The build application wrapper script
+        buildApp = pkgs.writeShellApplication {
+          name = "build";
+          runtimeInputs = [ pkgs.nodejs_22 pkgs.elixir emacsWithOxHugo ];
+          text = ''
+            NOTES_DIR="''${1:?Usage: build <path-to-notes-dir>}"
+            NOTES_DIR=$(realpath "$NOTES_DIR")
+            ORIG_CWD=$(pwd)
+
+            # Set up a writable working copy of the repo in a temp dir
+            WORK=$(mktemp -d)
+            trap 'rm -rf "$WORK"' EXIT
+            cp -r ${self}/. "$WORK/repo"
+            chmod -R u+w "$WORK/repo"
+
+            # Drop in pre-built node_modules
+            ln -s ${quartzDeps}/node_modules "$WORK/repo/node_modules"
+
+            # Drop in pre-fetched Mix deps so mix compile runs offline
+            cp -r ${pipelineMixDeps} "$WORK/repo/scripts/pipeline/deps"
+            chmod -R u+w "$WORK/repo/scripts/pipeline/deps"
+
+            # ox-hugo requires static/ to exist before it can copy image assets
+            mkdir -p "$WORK/repo/static"
+
+            # Run the export pipeline (org → md, citations transform)
+            NOTES_DIR="$NOTES_DIR" elixir "$WORK/repo/scripts/export.exs"
+
+            # Build the static site from within the repo copy so relative paths
+            # (e.g. ./package.json in constants.js) resolve correctly.
+            # --output is absolute so the result lands in the caller's cwd.
+            cd "$WORK/repo"
+            node quartz/bootstrap-cli.mjs build \
+              --directory "$WORK/repo/content" \
+              --output "$ORIG_CWD/public"
+          '';
+        };
+      in
+      {
+        devShells.default = pkgs.mkShell {
+          buildInputs = [
+            pkgs.nodejs_22
+            pkgs.elixir
+            emacsWithOxHugo
+            pkgs.mcp-nixos
+          ];
+
+          shellHook = ''
+            echo "Node $(node --version) / npm $(npm --version)"
+            elixir --version 2>/dev/null | head -1 || true
+            echo "Emacs $(emacs --version | head -1)"
+          '';
+        };
+
+        packages.default = buildApp;
+        packages.build = buildApp;
+
+        apps.default = { type = "app"; program = "${buildApp}/bin/build"; };
+        apps.build = { type = "app"; program = "${buildApp}/bin/build"; };
+      });
+}

notes/bus/emt-madrid.org

@@ -0,0 +1,16 @@
+:PROPERTIES:
+:ID:       emt-madrid
+:END:
+#+title: EMT Madrid (urban bus)
+
+Empresa Municipal de Transportes (EMT) operates the urban bus network
+within the municipality of Madrid — around 200 lines.
+
+* Notable lines
+- *Line 27* — connects Embajadores with Barrio de la Concepción, one of the
+  oldest routes in the network.
+- *Line 34* — Argüelles to Carabanchel, crossing the city centre via Gran Vía.
+- *Búho (owl) lines* — night buses running from Cibeles from midnight to 6 am.
+
+* See also
+- [[id:madrid-transport][Madrid Public Transport]]

notes/example-citation.org

@@ -0,0 +1,13 @@
+#+title: Example: Citation Reference
+
+This file demonstrates how org-citar citations pass through ox-hugo into
+markdown, where the pipeline transform resolves them.
+
+The methodology described in [cite:@podlovics2021journalArticle] provides a
+useful framework for analysis.
+
+Multiple citations can appear together:
+[cite:@podlovics2021journalArticle]
+
+Older bare-cite style (org-roam v1 / older citar) also works:
+cite:podlovics2021journalArticle

notes/example-images.org

@@ -0,0 +1,33 @@
+:PROPERTIES:
+:ID:       example-images
+:END:
+#+title: Example: Image References
+
+This note demonstrates the three image reference scenarios that the pipeline
+must handle.
+
+* Scenario 1: External image (URL)
+
+An image hosted on the web — ox-hugo passes the URL through as-is and no
+local file handling is needed.
+
+#+attr_html: :link "https://encrypted-tbn0.gstatic.com/images?q=tbn:ANd9GcSkzsTuLOt8esM6enoKwkzqA52G3p9hldlf2g&s"
+[[file:quartz-logo-external.png]]
+
+* Scenario 2: Local image (same notes directory)
+
+An image sitting next to this .org file inside the notes directory.
+ox-hugo copies files referenced with a relative path into the Hugo =static/=
+assets tree automatically.
+
+#+CAPTION: Quartz logo (local, same notes dir)
+[[file:quartz-logo.png]]
+
+* Scenario 3: External image (outside notes directory)
+
+An image that lives outside the notes directory entirely — for example a
+shared assets folder or a system path. ox-hugo still copies it into =static/=
+and rewrites the reference.
+
+#+CAPTION: Quartz logo (outside notes dir)
+[[file:../notes-external/external-location-image.png]]

notes/madrid-transport.org

@@ -0,0 +1,17 @@
+:PROPERTIES:
+:ID:       madrid-transport
+:END:
+#+title: Madrid Public Transport
+
+Madrid has one of the most extensive public transport networks in Europe,
+operated primarily by [[id:crtm][Consorcio Regional de Transportes de Madrid]] (CRTM).
+
+* Modes
+- [[id:metro-madrid][Metro de Madrid]] — 13 lines, ~300 km of track
+- [[id:emt-madrid][EMT Bus]] — urban buses within the city
+- Cercanías — suburban rail run by Renfe
+- Interurbano — regional buses to the wider Community of Madrid
+
+* Ticketing
+A single [[https://www.crtm.es][tarjeta transporte]] (transport card) works across all modes.
+The Multi card covers zones A–C and is topped up at any metro station.

notes/metro/metro-madrid.org

@@ -0,0 +1,18 @@
+:PROPERTIES:
+:ID:       metro-madrid
+:END:
+#+title: Metro de Madrid
+
+The Madrid Metro is the main rapid transit network in the city, opened in 1919.
+It is the second oldest metro in the Iberian Peninsula after Barcelona.
+
+* Key Lines
+| Line | Name            | Colour | Terminals                    |
+|------+-----------------+--------+------------------------------|
+| L1   | Pinar de Chamartín–Valdecarros | Blue   | Pinar de Chamartín / Valdecarros |
+| L6   | Circular        | Grey   | Circular (loop)              |
+| L10  | —               | Dark blue | Hospital Infanta Sofía / Tres Olivos |
+
+* See also
+- [[id:madrid-transport][Madrid Public Transport]]
+- [[id:sol-interchange][Sol interchange]]

notes/metro/sol-interchange.org

@@ -0,0 +1,12 @@
+:PROPERTIES:
+:ID:       sol-interchange
+:END:
+#+title: Sol (interchange)
+
+Sol is the busiest interchange station in the Madrid Metro, sitting beneath
+Puerta del Sol in the city centre.
+
+Lines serving Sol: [[id:metro-madrid][L1]], L2, L3.
+
+It also connects to the Cercanías hub underneath, making it the de-facto
+zero point of Madrid's public transport.

notes/roads/crtm.org

@@ -0,0 +1,22 @@
+:PROPERTIES:
+:ID:       crtm
+:END:
+#+title: CRTM — Consorcio Regional de Transportes de Madrid
+
+The CRTM is the regional authority that coordinates public transport across
+the Community of Madrid. It does not operate services directly but sets
+fares, zones, and integration policy.
+
+* Fare zones
+| Zone  | Coverage                    |
+|-------+-----------------------------|
+| A     | Municipality of Madrid      |
+| B1    | Inner ring municipalities   |
+| B2    | Outer ring municipalities   |
+| B3    | Further suburban area       |
+| C1–C2 | Commuter belt               |
+
+* Related
+- [[id:madrid-transport][Madrid Public Transport]]
+- [[id:metro-madrid][Metro de Madrid]]
+- [[id:emt-madrid][EMT Madrid]]

notes/roads/m30.org

@@ -0,0 +1,19 @@
+:PROPERTIES:
+:ID:       m30
+:END:
+#+title: M-30
+
+The M-30 is Madrid's innermost ring road, circling the city centre at a
+radius of roughly 3–5 km from Puerta del Sol.
+
+It runs mostly underground through the Madrid Río tunnel section along the
+Manzanares river, built during the 2004–2007 renovation that reclaimed the
+riverbank as a public park.
+
+* Key junctions
+- Nudo Norte — connects to A-1 (Burgos) and A-6 (La Coruña)
+- Nudo Sur — connects to A-4 (Cádiz) and A-42 (Toledo)
+
+* See also
+- [[id:crtm][CRTM]]
+- [[id:madrid-transport][Madrid Public Transport]]

opencode.json

@@ -0,0 +1,10 @@
+{
+  "$schema": "https://opencode.ai/config.json",
+  "mcp": {
+    "nixos": {
+      "type": "local",
+      "command": ["mcp-nixos"],
+      "enabled": true
+    }
+  }
+}

package.json

@@ -17,7 +17,10 @@
     "check": "tsc --noEmit && npx prettier . --check",
     "format": "npx prettier . --write",
     "test": "tsx --test",
-    "profile": "0x -D prof ./quartz/bootstrap-cli.mjs build --concurrency=1"
+    "profile": "0x -D prof ./quartz/bootstrap-cli.mjs build --concurrency=1",
+    "export": "elixir scripts/export.exs",
+    "build:notes": "elixir scripts/export.exs && npx quartz build",
+    "serve:notes": "elixir scripts/export.exs && npx quartz build --serve"
   },
   "engines": {
     "npm": ">=10.9.2",

quartz.config.ts

@@ -55,7 +55,7 @@ const config: QuartzConfig = {
   },
   plugins: {
     transformers: [
-      Plugin.FrontMatter(),
+      Plugin.FrontMatter({ delimiters: "+++", language: "toml" }),
       Plugin.CreatedModifiedDate({
         priority: ["frontmatter", "git", "filesystem"],
       }),
@@ -66,7 +66,11 @@ const config: QuartzConfig = {
         },
         keepBackground: false,
       }),
-      Plugin.ObsidianFlavoredMarkdown({ enableInHtmlEmbed: false }),
+      // OxHugoFlavouredMarkdown must come before GitHubFlavoredMarkdown.
+      // Note: not compatible with ObsidianFlavoredMarkdown — use one or the other.
+      // If ox-hugo exports TOML frontmatter, change FrontMatter to:
+      //   Plugin.FrontMatter({ delims: "+++", language: "toml" })
+      Plugin.OxHugoFlavouredMarkdown(),
       Plugin.GitHubFlavoredMarkdown(),
       Plugin.TableOfContents(),
       Plugin.CrawlLinks({ markdownLinkResolution: "shortest" }),

quartz/build.ts

@@ -71,7 +71,7 @@ async function buildQuartz(argv: Argv, mut: Mutex, clientRefresh: () => void) {
   console.log(`Cleaned output directory \`${output}\` in ${perf.timeSince("clean")}`)
 
   perf.addEvent("glob")
-  const allFiles = await glob("**/*.*", argv.directory, cfg.configuration.ignorePatterns)
+  const allFiles = await glob("**/*.*", argv.directory, cfg.configuration.ignorePatterns, false)
   const markdownPaths = allFiles.filter((fp) => fp.endsWith(".md")).sort()
   console.log(
     `Found ${markdownPaths.length} input files from \`${argv.directory}\` in ${perf.timeSince("glob")}`,

quartz/plugins/emitters/static.ts

@@ -7,6 +7,7 @@ import { dirname } from "path"
 export const Static: QuartzEmitterPlugin = () => ({
   name: "Static",
   async *emit({ argv, cfg }) {
+    // Copy Quartz's own internal static assets (quartz/static/) → output/static/
     const staticPath = joinSegments(QUARTZ, "static")
     const fps = await glob("**", staticPath, cfg.configuration.ignorePatterns)
     const outputStaticPath = joinSegments(argv.output, "static")
@@ -18,6 +19,21 @@ export const Static: QuartzEmitterPlugin = () => ({
       await fs.promises.copyFile(src, dest)
       yield dest
     }
+
+    // Copy user-facing static assets (static/) → output/ preserving paths.
+    // This mirrors Hugo's convention: static/ox-hugo/foo.png is served at /ox-hugo/foo.png,
+    // which matches the src="/ox-hugo/..." paths that ox-hugo writes into exported markdown.
+    const userStaticPath = "static"
+    if (fs.existsSync(userStaticPath)) {
+      const userFps = await glob("**", userStaticPath, cfg.configuration.ignorePatterns, false)
+      for (const fp of userFps) {
+        const src = joinSegments(userStaticPath, fp) as FilePath
+        const dest = joinSegments(argv.output, fp) as FilePath
+        await fs.promises.mkdir(dirname(dest), { recursive: true })
+        await fs.promises.copyFile(src, dest)
+        yield dest
+      }
+    }
   },
   async *partialEmit() {},
 })

quartz/plugins/transformers/oxhugofm.ts

@@ -27,7 +27,10 @@ const defaultOptions: Options = {
 const relrefRegex = new RegExp(/\[([^\]]+)\]\(\{\{< relref "([^"]+)" >\}\}\)/, "g")
 const predefinedHeadingIdRegex = new RegExp(/(.*) {#(?:.*)}/, "g")
 const hugoShortcodeRegex = new RegExp(/{{(.*)}}/, "g")
-const figureTagRegex = new RegExp(/< ?figure src="(.*)" ?>/, "g")
+// Matches the full Hugo {{< figure src="..." ... >}} shortcode and captures src.
+// Must run before the generic shortcode stripper to avoid partial-match issues
+// with captions that contain HTML (e.g. <span class="figure-number">).
+const figureShortcodeRegex = new RegExp(/{{<\s*figure\b[^}]*\bsrc="([^"]*)"[^}]*>}}/, "g")
 // \\\\\( -> matches \\(
 // (.+?) -> Lazy match for capturing the equation
 // \\\\\) -> matches \\)
@@ -70,6 +73,14 @@ export const OxHugoFlavouredMarkdown: QuartzTransformerPlugin<Partial<Options>>
         })
       }
 
+      if (opts.replaceFigureWithMdImg) {
+        src = src.toString()
+        src = src.replaceAll(figureShortcodeRegex, (_value, ...capture) => {
+          const [imgSrc] = capture
+          return `![](${imgSrc})`
+        })
+      }
+
       if (opts.removeHugoShortcode) {
         src = src.toString()
         src = src.replaceAll(hugoShortcodeRegex, (_value, ...capture) => {
@@ -78,14 +89,6 @@ export const OxHugoFlavouredMarkdown: QuartzTransformerPlugin<Partial<Options>>
         })
       }
 
-      if (opts.replaceFigureWithMdImg) {
-        src = src.toString()
-        src = src.replaceAll(figureTagRegex, (_value, ...capture) => {
-          const [src] = capture
-          return `![](${src})`
-        })
-      }
-
       if (opts.replaceOrgLatex) {
         src = src.toString()
         src = src.replaceAll(inlineLatexRegex, (_value, ...capture) => {

quartz/util/glob.ts

@@ -10,12 +10,13 @@ export async function glob(
   pattern: string,
   cwd: string,
   ignorePatterns: string[],
+  respectGitignore: boolean = true,
 ): Promise<FilePath[]> {
   const fps = (
     await globby(pattern, {
       cwd,
       ignore: ignorePatterns,
-      gitignore: true,
+      gitignore: respectGitignore,
     })
   ).map(toPosixPath)
   return fps as FilePath[]

scripts/export.exs

@@ -0,0 +1,217 @@
+#!/usr/bin/env elixir
+# Export org-roam notes (per-file) to content/ via ox-hugo,
+# then run the markdown transformation pipeline (citations, etc.).
+#
+# Usage:
+#   NOTES_DIR=~/notes elixir scripts/export.exs
+#   elixir scripts/export.exs /path/to/notes
+#
+# Optional env vars:
+#   BIBTEX_FILE   — path to a .bib file used as citation fallback
+#   ZOTERO_URL    — Zotero Better BibTeX base URL (default: http://localhost:23119)
+#   CITATION_MODE — silent | warn (default) | strict
+#
+# The positional argument takes precedence over the NOTES_DIR env var.
+
+# ---------------------------------------------------------------------------
+# Load the pipeline Mix project so its modules are available in this script.
+# ---------------------------------------------------------------------------
+repo_root = __DIR__ |> Path.join("..") |> Path.expand()
+pipeline_dir = Path.join(repo_root, "scripts/pipeline")
+
+# Compile and load the pipeline project's modules into this runtime.
+# Mix.install is NOT used here because we have a local Mix project — instead
+# we compile it and push its beam files onto the code path.
+#
+# This runs `mix deps.get` + `mix compile` the first time; subsequent runs
+# use the compiled artifacts from _build/ (fast, same as Mix caching).
+{_, 0} =
+  System.cmd("mix", ["deps.get", "--quiet"],
+    cd: pipeline_dir,
+    env: [{"MIX_ENV", "prod"}],
+    into: IO.stream()
+  )
+
+{_, 0} =
+  System.cmd("mix", ["compile", "--quiet"],
+    cd: pipeline_dir,
+    env: [{"MIX_ENV", "prod"}],
+    into: IO.stream()
+  )
+
+# Add compiled beam files to the load path so we can call pipeline modules.
+pipeline_build = Path.join(pipeline_dir, "_build/prod/lib")
+
+pipeline_build
+|> File.ls!()
+|> Enum.each(fn app ->
+  ebin = Path.join([pipeline_build, app, "ebin"])
+  if File.dir?(ebin), do: Code.prepend_path(ebin)
+end)
+
+# Start the pipeline OTP application (which starts Finch for HTTP).
+Application.ensure_all_started(:pipeline)
+
+# ---------------------------------------------------------------------------
+# Argument / env resolution
+# ---------------------------------------------------------------------------
+
+notes_dir =
+  case System.argv() do
+    [dir | _] -> dir
+    [] ->
+      System.get_env("NOTES_DIR") ||
+        (IO.puts(:stderr, "Usage: NOTES_DIR=/path/to/notes elixir scripts/export.exs")
+         System.halt(1))
+  end
+
+notes_dir = Path.expand(notes_dir)
+content_dir = Path.join(repo_root, "content")
+
+unless File.dir?(notes_dir) do
+  IO.puts(:stderr, "Error: notes directory does not exist: #{notes_dir}")
+  System.halt(1)
+end
+
+# ---------------------------------------------------------------------------
+# Phase 1: Wipe content/
+# ---------------------------------------------------------------------------
+IO.puts("==> Wiping #{content_dir}")
+
+content_dir
+|> File.ls!()
+|> Enum.reject(&(&1 == ".gitkeep"))
+|> Enum.each(fn entry ->
+  Path.join(content_dir, entry) |> File.rm_rf!()
+end)
+
+# ---------------------------------------------------------------------------
+# Phase 2: Export org files via Emacs + ox-hugo
+# ---------------------------------------------------------------------------
+IO.puts("==> Exporting org files from #{notes_dir}")
+
+org_files =
+  Path.join(notes_dir, "**/*.org")
+  |> Path.wildcard()
+
+if org_files == [] do
+  IO.puts("No .org files found in #{notes_dir}")
+  System.halt(0)
+end
+
+results =
+  Enum.map(org_files, fn orgfile ->
+    IO.puts("  exporting: #{orgfile}")
+
+    section =
+      orgfile
+      |> Path.dirname()
+      |> Path.relative_to(notes_dir)
+
+    {output, exit_code} =
+      System.cmd(
+        "emacs",
+        [
+          "--batch",
+          "--eval", "(require 'ox-hugo)",
+          "--eval", ~s[(setq org-hugo-base-dir "#{repo_root}")],
+          "--eval", ~s[(setq org-hugo-default-section-directory "#{section}")],
+          "--visit", orgfile,
+          "--funcall", "org-hugo-export-to-md"
+        ],
+        stderr_to_stdout: true
+      )
+
+    filtered =
+      output
+      |> String.split("\n")
+      |> Enum.reject(&String.match?(&1, ~r/^Loading|^ad-handle|^For information/))
+      |> Enum.join("\n")
+
+    if filtered != "", do: IO.puts(filtered)
+
+    {orgfile, exit_code}
+  end)
+
+failures = Enum.filter(results, fn {_, code} -> code != 0 end)
+
+if failures != [] do
+  IO.puts(:stderr, "\nFailed to export #{length(failures)} file(s):")
+  Enum.each(failures, fn {f, code} -> IO.puts(:stderr, "  [exit #{code}] #{f}") end)
+  System.halt(1)
+end
+
+# ---------------------------------------------------------------------------
+# Phase 3: Markdown transformation pipeline
+# ---------------------------------------------------------------------------
+IO.puts("==> Running markdown pipeline")
+
+pipeline_opts = %{
+  zotero_url: System.get_env("ZOTERO_URL", "http://localhost:23119"),
+  bibtex_file: System.get_env("BIBTEX_FILE"),
+  citation_mode:
+    case System.get_env("CITATION_MODE", "warn") do
+      "silent" -> :silent
+      "strict" -> :strict
+      _ -> :warn
+    end
+}
+
+transforms = [Pipeline.Transforms.Citations]
+
+case Pipeline.run(content_dir, transforms, pipeline_opts) do
+  {:ok, stats} ->
+    Enum.each(stats, fn {mod, count} ->
+      IO.puts("  #{inspect(mod)}: #{count} file(s) modified")
+    end)
+
+  {:error, reason} ->
+    IO.puts(:stderr, "Pipeline error: #{inspect(reason)}")
+    System.halt(1)
+end
+
+# ---------------------------------------------------------------------------
+# Phase 4: Generate default index.md if none was exported
+# ---------------------------------------------------------------------------
+md_count =
+  Path.join(content_dir, "**/*.md")
+  |> Path.wildcard()
+  |> length()
+
+index_path = Path.join(content_dir, "index.md")
+
+unless File.exists?(index_path) do
+  IO.puts("==> Generating default index.md")
+
+  pages =
+    Path.join(content_dir, "**/*.md")
+    |> Path.wildcard()
+    |> Enum.map(fn path ->
+      slug = Path.relative_to(path, content_dir) |> Path.rootname()
+
+      title =
+        path
+        |> File.read!()
+        |> then(fn content ->
+          case Regex.run(~r/^title\s*=\s*"(.+)"/m, content) do
+            [_, t] -> t
+            _ -> slug
+          end
+        end)
+
+      {slug, title}
+    end)
+    |> Enum.sort_by(fn {_, title} -> title end)
+    |> Enum.map(fn {slug, title} -> "- [#{title}](#{slug})" end)
+    |> Enum.join("\n")
+
+  File.write!(index_path, """
+  ---
+  title: Index
+  ---
+
+  #{pages}
+  """)
+end
+
+IO.puts("==> Done. #{md_count} markdown files in #{content_dir}")

scripts/pipeline/lib/pipeline.ex

@@ -0,0 +1,83 @@
+defmodule Pipeline do
+  @moduledoc """
+  Post-export markdown transformation pipeline.
+
+  Applies a list of transform modules sequentially over every .md file
+  in a content directory. Each transform module must implement:
+
+      apply(content :: String.t(), opts :: map()) :: String.t()
+
+  Transforms are applied in the order given. A file is rewritten only
+  when at least one transform mutates its content (checked via equality).
+
+  ## Usage
+
+      opts = %{
+        zotero_url: "http://localhost:23119",
+        bibtex_file: System.get_env("BIBTEX_FILE"),
+        citation_mode: :warn   # :silent | :warn | :strict
+      }
+
+      Pipeline.run(content_dir, [Pipeline.Transforms.Citations], opts)
+  """
+
+  require Logger
+
+  @type transform :: module()
+  @type opts :: map()
+
+  @doc """
+  Run all transforms over every .md file under `content_dir`.
+  Returns `{:ok, stats}` where stats maps each transform to a count of files it changed.
+  """
+  @spec run(String.t(), [transform()], opts()) :: {:ok, map()}
+  def run(content_dir, transforms, opts \\ %{}) do
+    md_files =
+      content_dir
+      |> Path.join("**/*.md")
+      |> Path.wildcard()
+
+    if md_files == [] do
+      Logger.warning("Pipeline: no .md files found in #{content_dir}")
+      {:ok, %{}}
+    else
+      Logger.info("Pipeline: processing #{length(md_files)} markdown files with #{length(transforms)} transform(s)")
+
+      # Initialise transforms (allows them to perform setup such as loading a .bib file).
+      # Each transform module must implement the Pipeline.Transform behaviour.
+      initialized =
+        Enum.map(transforms, fn mod ->
+          state = mod.init(opts)
+          {mod, state}
+        end)
+
+      stats =
+        Enum.reduce(md_files, %{}, fn path, acc ->
+          original = File.read!(path)
+
+          {transformed, file_stats} =
+            Enum.reduce(initialized, {original, %{}}, fn {mod, state}, {content, fstats} ->
+              result = mod.apply(content, state, opts)
+              changed = result != content
+              {result, Map.update(fstats, mod, (if changed, do: 1, else: 0), &(&1 + (if changed, do: 1, else: 0)))}
+            end)
+
+          if transformed != original do
+            File.write!(path, transformed)
+            Logger.debug("Pipeline: updated #{Path.relative_to_cwd(path)}")
+          end
+
+          Map.merge(acc, file_stats, fn _k, a, b -> a + b end)
+        end)
+
+      Enum.each(initialized, fn {mod, state} ->
+        # teardown/1 is optional in the behaviour
+        if function_exported?(mod, :teardown, 1) do
+          mod.teardown(state)
+        end
+      end)
+
+      {:ok, stats}
+    end
+  end
+end

scripts/pipeline/lib/pipeline/application.ex

@@ -0,0 +1,14 @@
+defmodule Pipeline.Application do
+  @moduledoc false
+  use Application
+
+  @impl true
+  def start(_type, _args) do
+    children = [
+      {Finch, name: Pipeline.Finch}
+    ]
+
+    opts = [strategy: :one_for_one, name: Pipeline.Supervisor]
+    Supervisor.start_link(children, opts)
+  end
+end

scripts/pipeline/lib/pipeline/resolvers/bibtex.ex

@@ -0,0 +1,178 @@
+defmodule Pipeline.Resolvers.BibTeX do
+  @moduledoc """
+  Resolves citation keys from a local BibTeX (.bib) file.
+
+  Configured via the `BIBTEX_FILE` environment variable, or passed directly
+  as `opts.bibtex_file`. The file is parsed once at init time and the
+  resulting entry map is reused for all lookups.
+
+  Supports extracting: author last names, year, title, DOI, URL.
+
+  BibTeX entry format parsed:
+
+      @type{citationkey,
+        author  = {Last, First and Last2, First2},
+        year    = {2021},
+        title   = {Some Title},
+        doi     = {10.xxxx/yyyy},
+        url     = {https://example.com},
+      }
+
+  Returns `{:ok, %{label: "Author, Year", url: "..."}}` or `:error`.
+  """
+
+  require Logger
+
+  # ------------------------------------------------------------------
+  # Public API
+  # ------------------------------------------------------------------
+
+  @doc """
+  Parse a .bib file and return a map of `%{citation_key => entry_map}`.
+  Returns `{:ok, entries}` or `{:error, reason}`.
+  """
+  @spec load(String.t()) :: {:ok, map()} | {:error, term()}
+  def load(path) do
+    case File.read(path) do
+      {:ok, content} ->
+        entries = parse_entries(content)
+        Logger.info("BibTeX: loaded #{map_size(entries)} entries from #{path}")
+        {:ok, entries}
+
+      {:error, reason} ->
+        {:error, reason}
+    end
+  end
+
+  @doc """
+  Resolve a citation key from pre-loaded BibTeX entries.
+  """
+  @spec resolve(String.t(), map()) :: {:ok, map()} | :error
+  def resolve(key, entries) do
+    case Map.fetch(entries, key) do
+      {:ok, entry} ->
+        label = build_label(entry)
+        url = build_url(entry)
+        {:ok, %{label: label, url: url}}
+
+      :error ->
+        :error
+    end
+  end
+
+  # ------------------------------------------------------------------
+  # Parsing
+  # ------------------------------------------------------------------
+
+  # Match @type{key, ...fields...}
+  # We handle nested braces by scanning character by character after
+  # finding the opening, rather than relying on a single regex.
+  @entry_header ~r/@\w+\s*\{\s*([^,\s]+)\s*,/
+
+  defp parse_entries(content) do
+    # Split on "@" boundaries, then parse each chunk
+    content
+    |> String.split(~r/(?=@\w+\s*\{)/, trim: true)
+    |> Enum.reduce(%{}, fn chunk, acc ->
+      case Regex.run(@entry_header, chunk) do
+        [_, key] ->
+          fields = parse_fields(chunk)
+          Map.put(acc, String.trim(key), fields)
+
+        _ ->
+          acc
+      end
+    end)
+  end
+
+  # Extract key = {value} or key = "value" pairs from an entry block.
+  # Handles simple single-depth braces; good enough for common fields.
+  @field_regex ~r/(\w+)\s*=\s*(?:\{([^{}]*(?:\{[^{}]*\}[^{}]*)*)\}|"([^"]*)")/
+
+  defp parse_fields(chunk) do
+    @field_regex
+    |> Regex.scan(chunk)
+    |> Enum.reduce(%{}, fn match, acc ->
+      field_name = Enum.at(match, 1) |> String.downcase()
+      # Value is in capture group 2 (braces) or 3 (quotes)
+      value =
+        case {Enum.at(match, 2, ""), Enum.at(match, 3, "")} do
+          {"", q} -> q
+          {b, _} -> b
+        end
+
+      Map.put(acc, field_name, String.trim(value))
+    end)
+  end
+
+  # ------------------------------------------------------------------
+  # Label & URL building
+  # ------------------------------------------------------------------
+
+  defp build_label(entry) do
+    author_part =
+      entry
+      |> Map.get("author", "")
+      |> parse_authors()
+      |> format_authors()
+
+    year = Map.get(entry, "year", Map.get(entry, "date", ""))
+    year = extract_year(year)
+
+    if year && author_part != "", do: "#{author_part}, #{year}", else: author_part
+  end
+
+  defp parse_authors(""), do: []
+
+  defp parse_authors(author_str) do
+    author_str
+    |> String.split(" and ", trim: true)
+    |> Enum.map(&extract_last_name/1)
+    |> Enum.reject(&(&1 == ""))
+  end
+
+  # Handles "Last, First" and "First Last" formats
+  defp extract_last_name(name) do
+    name = String.trim(name)
+
+    cond do
+      String.contains?(name, ",") ->
+        name |> String.split(",") |> List.first() |> String.trim()
+
+      String.contains?(name, " ") ->
+        name |> String.split(" ") |> List.last() |> String.trim()
+
+      true ->
+        name
+    end
+  end
+
+  defp format_authors([]), do: "Unknown"
+  defp format_authors([single]), do: single
+  defp format_authors([first | rest]), do: "#{first} & #{List.last(rest)}"
+
+  defp extract_year(""), do: nil
+
+  defp extract_year(str) do
+    case Regex.run(~r/\b(\d{4})\b/, str) do
+      [_, year] -> year
+      _ -> nil
+    end
+  end
+
+  defp build_url(entry) do
+    cond do
+      doi = Map.get(entry, "doi", "") |> non_empty() ->
+        "https://doi.org/#{doi}"
+
+      url = Map.get(entry, "url", "") |> non_empty() ->
+        url
+
+      true ->
+        nil
+    end
+  end
+
+  defp non_empty(""), do: nil
+  defp non_empty(v), do: v
+end

scripts/pipeline/lib/pipeline/resolvers/doi.ex

@@ -0,0 +1,18 @@
+defmodule Pipeline.Resolvers.DOI do
+  @moduledoc """
+  Last-resort citation resolver — always succeeds.
+
+  If the citation key looks like a DOI (starts with "10."), returns a
+  `https://doi.org/...` link. Otherwise returns the key itself as a
+  plain label with no URL.
+  """
+
+  @spec resolve(String.t()) :: {:ok, map()}
+  def resolve(key) do
+    if String.starts_with?(key, "10.") do
+      {:ok, %{label: key, url: "https://doi.org/#{key}"}}
+    else
+      {:ok, %{label: key, url: nil}}
+    end
+  end
+end

scripts/pipeline/lib/pipeline/resolvers/zotero.ex

@@ -0,0 +1,182 @@
+defmodule Pipeline.Resolvers.Zotero do
+  @moduledoc """
+  Resolves citation keys via Zotero Better BibTeX's JSON-RPC API.
+
+  Requires Zotero to be running with the Better BibTeX plugin installed.
+  Default endpoint: http://localhost:23119/better-bibtex/json-rpc
+
+  Resolution strategy:
+    1. Search by citation key via `item.search`
+    2. If found, try to get a PDF attachment link (zotero://open-pdf/...)
+    3. Fall back to zotero://select/items/@key
+
+  Returns `{:ok, %{label: "Author, Year", url: "zotero://..."}}` or `:error`.
+  """
+
+  require Logger
+
+  @rpc_path "/better-bibtex/json-rpc"
+
+  @doc """
+  Attempt to resolve `key` against a running Zotero instance.
+  `base_url` defaults to `http://localhost:23119`.
+  """
+  @spec resolve(String.t(), String.t()) :: {:ok, map()} | :error
+  def resolve(key, base_url \\ "http://localhost:23119") do
+    url = base_url <> @rpc_path
+
+    payload =
+      Jason.encode!(%{
+        jsonrpc: "2.0",
+        method: "item.search",
+        params: [
+          [["citationKey", "is", key]]
+        ],
+        id: 1
+      })
+
+    case Req.post(url,
+           body: payload,
+           headers: [{"content-type", "application/json"}],
+           receive_timeout: 5_000,
+           finch: Pipeline.Finch
+         ) do
+      {:ok, %{status: 200, body: body}} ->
+        parse_response(body, key, base_url)
+
+      {:ok, %{status: status}} ->
+        Logger.debug("Zotero: unexpected HTTP #{status} for key #{key}")
+        :error
+
+      {:error, reason} ->
+        Logger.debug("Zotero: connection failed for key #{key}: #{inspect(reason)}")
+        :error
+
+      other ->
+        Logger.debug("Zotero: unexpected result for key #{key}: #{inspect(other)}")
+        :error
+    end
+  rescue
+    e ->
+      Logger.debug("Zotero: exception resolving key #{key}: #{inspect(e)}")
+      :error
+  end
+
+  # ------------------------------------------------------------------
+  # Private helpers
+  # ------------------------------------------------------------------
+
+  defp parse_response(%{"result" => [item | _]}, key, base_url) do
+    label = build_label(item)
+    url = resolve_url(item, key, base_url)
+    {:ok, %{label: label, url: url}}
+  end
+
+  defp parse_response(%{"result" => []}, key, _base_url) do
+    Logger.debug("Zotero: no item found for key #{key}")
+    :error
+  end
+
+  defp parse_response(%{"error" => err}, key, _base_url) do
+    Logger.debug("Zotero: RPC error for key #{key}: #{inspect(err)}")
+    :error
+  end
+
+  defp parse_response(body, key, _base_url) do
+    Logger.debug("Zotero: unexpected response shape for key #{key}: #{inspect(body)}")
+    :error
+  end
+
+  defp fetch_pdf_url(key, base_url) do
+    payload =
+      Jason.encode!(%{
+        jsonrpc: "2.0",
+        method: "item.attachments",
+        params: [key],
+        id: 2
+      })
+
+    case Req.post(base_url <> @rpc_path,
+           body: payload,
+           headers: [{"content-type", "application/json"}],
+           receive_timeout: 5_000,
+           finch: Pipeline.Finch
+         ) do
+      {:ok, %{status: 200, body: %{"result" => attachments}}} when is_list(attachments) ->
+        attachments
+        |> Enum.find_value(fn att ->
+          open = Map.get(att, "open", "")
+          path = Map.get(att, "path", "")
+          if String.ends_with?(path, ".pdf"), do: open, else: nil
+        end)
+
+      _ ->
+        nil
+    end
+  rescue
+    _ -> nil
+  end
+
+  # CSL-JSON format: authors are under "author" with "family"/"given" keys.
+  # Year is under "issued" -> "date-parts" -> [[year, month, day]].
+  defp build_label(item) do
+    authors = Map.get(item, "author", [])
+    year = extract_year(item)
+
+    author_part =
+      case authors do
+        [] ->
+          "Unknown"
+
+        [single] ->
+          Map.get(single, "family", Map.get(single, "literal", "Unknown"))
+
+        [first | rest] ->
+          first_name = Map.get(first, "family", Map.get(first, "literal", "Unknown"))
+          last_name =
+            rest
+            |> List.last()
+            |> then(&Map.get(&1, "family", Map.get(&1, "literal", "Unknown")))
+
+          "#{first_name} & #{last_name}"
+      end
+
+    if year, do: "#{author_part}, #{year}", else: author_part
+  end
+
+  # "issued": {"date-parts": [["2021", 2, 3]]}
+  defp extract_year(item) do
+    case get_in(item, ["issued", "date-parts"]) do
+      [[year | _] | _] -> to_string(year)
+      _ -> nil
+    end
+  end
+
+  defp resolve_url(item, key, base_url) do
+    # Prefer zotero://open-pdf/... for items with a PDF attachment.
+    # Fall back to zotero://select/library/items/KEY to open the item in Zotero.
+    # The "id" field is a URI like "http://zotero.org/users/123/items/ABCD1234".
+    pdf_url = fetch_pdf_url(key, base_url)
+
+    if pdf_url do
+      pdf_url
+    else
+      item_key =
+        item
+        |> Map.get("id", "")
+        |> String.split("/")
+        |> List.last()
+        |> non_empty()
+
+      if item_key do
+        "zotero://select/library/items/#{item_key}"
+      else
+        "zotero://select/items/@#{key}"
+      end
+    end
+  end
+
+  defp non_empty(nil), do: nil
+  defp non_empty(""), do: nil
+  defp non_empty(v), do: v
+end

scripts/pipeline/lib/pipeline/transform.ex

@@ -0,0 +1,48 @@
+defmodule Pipeline.Transform do
+  @moduledoc """
+  Behaviour that all markdown transform modules must implement.
+
+  ## Callbacks
+
+  - `init/1`      — called once before processing; returns transform-specific state.
+                    Default implementation returns the opts map unchanged.
+  - `apply/3`     — called per .md file; returns the (possibly modified) content.
+  - `teardown/1`  — optional cleanup after all files are processed.
+
+  ## Example
+
+      defmodule MyTransform do
+        @behaviour Pipeline.Transform
+
+        @impl true
+        def init(opts), do: %{some_state: opts[:value]}
+
+        @impl true
+        def apply(content, state, _opts) do
+          String.replace(content, "foo", state.some_state)
+        end
+      end
+  """
+
+  @doc "One-time initialisation. Returns opaque state passed to apply/3."
+  @callback init(opts :: map()) :: term()
+
+  @doc "Transform file content. Returns the (possibly modified) content string."
+  @callback apply(content :: String.t(), state :: term(), opts :: map()) :: String.t()
+
+  @doc "Optional cleanup after all files are processed."
+  @callback teardown(state :: term()) :: :ok
+
+  @optional_callbacks teardown: 1
+
+  defmacro __using__(_) do
+    quote do
+      @behaviour Pipeline.Transform
+
+      @impl Pipeline.Transform
+      def init(opts), do: opts
+
+      defoverridable init: 1
+    end
+  end
+end

scripts/pipeline/lib/pipeline/transforms/citations.ex

@@ -0,0 +1,231 @@
+defmodule Pipeline.Transforms.Citations do
+  @moduledoc """
+  Markdown transform: resolves org-citar citation keys to hyperlinks.
+
+  ## Recognised citation syntax (as output by ox-hugo from org-citar)
+
+      [cite:@key]           → org-cite / citar standard (most common)
+      [cite:@key1;@key2]    → multiple citations
+      cite:key              → older roam-style bare cite syntax
+
+  ## Resolution chain (in order)
+
+    1. Zotero (live instance via Better BibTeX JSON-RPC)   — preferred
+    2. BibTeX file (BIBTEX_FILE env var)                   — fallback
+    3. DOI / bare key                                       — always succeeds
+
+  ## Modes (opts.citation_mode)
+
+    :silent  — silently use DOI/bare-key fallback when Zotero+BibTeX fail
+    :warn    — (default) emit a Logger.warning for unresolved keys
+    :strict  — raise on unresolved keys (aborts pipeline)
+
+  ## Format
+
+  Resolved citations are rendered as:
+
+      [Label](url)       when a URL is available
+      [Label]            when no URL could be determined (bare key fallback)
+
+  Multiple semicolon-separated keys become space-separated links:
+
+      [cite:@a;@b]  →  [Author A, 2020](url_a) [Author B, 2019](url_b)
+
+  ## init/1 callback
+
+  Loads the BibTeX file (if configured) once before processing begins,
+  and probes Zotero availability, emitting warnings as appropriate.
+  """
+
+  @behaviour Pipeline.Transform
+
+  require Logger
+
+  alias Pipeline.Resolvers.Zotero
+  alias Pipeline.Resolvers.BibTeX
+  alias Pipeline.Resolvers.DOI
+
+  # Match [cite:@key] and [cite:@key1;@key2;...] (org-cite / citar style)
+  @cite_bracket_regex ~r/\[cite:(@[^\]]+)\]/
+
+  # Match bare cite:key (older roam style, no brackets, no @ prefix)
+  @cite_bare_regex ~r/(?<![(\[])cite:([a-zA-Z0-9_:-]+)/
+
+  # ------------------------------------------------------------------
+  # Pipeline callbacks
+  # ------------------------------------------------------------------
+
+  @doc """
+  Called once before processing any files. Loads BibTeX, probes Zotero.
+  Returns a state map passed to every `apply/3` call.
+  """
+  def init(opts) do
+    bibtex_entries = load_bibtex(opts)
+    zotero_available = probe_zotero(opts)
+
+    if not zotero_available and bibtex_entries == %{} do
+      Logger.warning(
+        "Citations: neither Zotero nor a BibTeX file is available. " <>
+          "All citations will fall back to bare-key rendering. " <>
+          "Set BIBTEX_FILE env var or start Zotero with Better BibTeX to resolve citations."
+      )
+    end
+
+    %{
+      bibtex_entries: bibtex_entries,
+      zotero_available: zotero_available,
+      zotero_url: Map.get(opts, :zotero_url, "http://localhost:23119"),
+      citation_mode: Map.get(opts, :citation_mode, :warn)
+    }
+  end
+
+  @doc """
+  Apply citation resolution to a single markdown file's content.
+  """
+  def apply(content, state, _opts) do
+    content
+    |> resolve_bracket_citations(state)
+    |> resolve_bare_citations(state)
+  end
+
+  # ------------------------------------------------------------------
+  # Resolution passes
+  # ------------------------------------------------------------------
+
+  defp resolve_bracket_citations(content, state) do
+    Regex.replace(@cite_bracket_regex, content, fn _full, keys_str ->
+      keys_str
+      |> String.split(";")
+      |> Enum.map(&String.trim/1)
+      |> Enum.map(fn "@" <> key -> key end)
+      |> Enum.map(&resolve_key(&1, state))
+      |> Enum.join(" ")
+    end)
+  end
+
+  defp resolve_bare_citations(content, state) do
+    Regex.replace(@cite_bare_regex, content, fn _full, key ->
+      resolve_key(key, state)
+    end)
+  end
+
+  # ------------------------------------------------------------------
+  # Single-key resolution chain
+  # ------------------------------------------------------------------
+
+  defp resolve_key(key, state) do
+    info =
+      with :error <- try_zotero(key, state),
+           :error <- try_bibtex(key, state) do
+        handle_unresolved(key, state)
+      else
+        {:ok, citation_info} -> citation_info
+      end
+
+    format_result(info)
+  end
+
+  defp try_zotero(_key, %{zotero_available: false}), do: :error
+
+  defp try_zotero(key, %{zotero_url: url}) do
+    Zotero.resolve(key, url)
+  end
+
+  defp try_bibtex(_key, %{bibtex_entries: entries}) when map_size(entries) == 0, do: :error
+
+  defp try_bibtex(key, %{bibtex_entries: entries}) do
+    BibTeX.resolve(key, entries)
+  end
+
+  defp handle_unresolved(key, %{citation_mode: mode}) do
+    case mode do
+      :strict ->
+        raise "Citations: could not resolve citation key '#{key}' and mode is :strict"
+
+      :warn ->
+        Logger.warning("Citations: unresolved citation key '#{key}' — using bare-key fallback")
+        {:ok, result} = DOI.resolve(key)
+        result
+
+      :silent ->
+        {:ok, result} = DOI.resolve(key)
+        result
+    end
+  end
+
+  defp format_result(%{label: label, url: nil}), do: "[#{label}]"
+  defp format_result(%{label: label, url: url}), do: "[#{label}](#{url})"
+
+  # ------------------------------------------------------------------
+  # Init helpers
+  # ------------------------------------------------------------------
+
+  defp load_bibtex(opts) do
+    path = Map.get(opts, :bibtex_file) || System.get_env("BIBTEX_FILE")
+
+    cond do
+      is_nil(path) ->
+        Logger.debug("Citations: BIBTEX_FILE not set — BibTeX resolver disabled")
+        %{}
+
+      not File.exists?(path) ->
+        Logger.warning("Citations: BIBTEX_FILE=#{path} does not exist — BibTeX resolver disabled")
+        %{}
+
+      true ->
+        case BibTeX.load(path) do
+          {:ok, entries} -> entries
+          {:error, reason} ->
+            Logger.warning("Citations: failed to load BibTeX file #{path}: #{inspect(reason)}")
+            %{}
+        end
+    end
+  end
+
+  defp probe_zotero(opts) do
+    url = Map.get(opts, :zotero_url, "http://localhost:23119")
+
+    # Use a no-op JSON-RPC call to probe availability.
+    # /better-bibtex/cayw is intentionally avoided — it blocks waiting for
+    # user interaction and never returns without a pick.
+    payload =
+      Jason.encode!(%{
+        jsonrpc: "2.0",
+        method: "item.search",
+        params: [[[]]],
+        id: 0
+      })
+
+    result =
+      try do
+        Req.post(url <> "/better-bibtex/json-rpc",
+          body: payload,
+          headers: [{"content-type", "application/json"}],
+          receive_timeout: 3_000,
+          finch: Pipeline.Finch
+        )
+      rescue
+        e -> {:error, e}
+      end
+
+    case result do
+      {:ok, %{status: 200}} ->
+        Logger.info("Citations: Zotero Better BibTeX is available at #{url}")
+        true
+
+      {:ok, %{status: status}} ->
+        Logger.warning(
+          "Citations: Zotero responded HTTP #{status} at #{url} — " <>
+            "is Better BibTeX installed?"
+        )
+        false
+
+      _ ->
+        Logger.warning(
+          "Citations: Zotero not reachable at #{url} — " <>
+            "start Zotero with Better BibTeX or set BIBTEX_FILE as fallback"
+        )
+        false
+    end
+  end
+end

scripts/pipeline/mix.exs

@@ -0,0 +1,27 @@
+defmodule Pipeline.MixProject do
+  use Mix.Project
+
+  def project do
+    [
+      app: :pipeline,
+      version: "0.1.0",
+      elixir: "~> 1.15",
+      start_permanent: Mix.env() == :prod,
+      deps: deps()
+    ]
+  end
+
+  def application do
+    [
+      extra_applications: [:logger, :inets, :ssl],
+      mod: {Pipeline.Application, []}
+    ]
+  end
+
+  defp deps do
+    [
+      {:req, "~> 0.5"},
+      {:jason, "~> 1.4"}
+    ]
+  end
+end

scripts/pipeline/mix.lock

@@ -0,0 +1,11 @@
+%{
+  "finch": {:hex, :finch, "0.21.0", "b1c3b2d48af02d0c66d2a9ebfb5622be5c5ecd62937cf79a88a7f98d48a8290c", [:mix], [{:mime, "~> 1.0 or ~> 2.0", [hex: :mime, repo: "hexpm", optional: false]}, {:mint, "~> 1.6.2 or ~> 1.7", [hex: :mint, repo: "hexpm", optional: false]}, {:nimble_options, "~> 0.4 or ~> 1.0", [hex: :nimble_options, repo: "hexpm", optional: false]}, {:nimble_pool, "~> 1.1", [hex: :nimble_pool, repo: "hexpm", optional: false]}, {:telemetry, "~> 0.4 or ~> 1.0", [hex: :telemetry, repo: "hexpm", optional: false]}], "hexpm", "87dc6e169794cb2570f75841a19da99cfde834249568f2a5b121b809588a4377"},
+  "hpax": {:hex, :hpax, "1.0.3", "ed67ef51ad4df91e75cc6a1494f851850c0bd98ebc0be6e81b026e765ee535aa", [:mix], [], "hexpm", "8eab6e1cfa8d5918c2ce4ba43588e894af35dbd8e91e6e55c817bca5847df34a"},
+  "jason": {:hex, :jason, "1.4.4", "b9226785a9aa77b6857ca22832cffa5d5011a667207eb2a0ad56adb5db443b8a", [:mix], [{:decimal, "~> 1.0 or ~> 2.0", [hex: :decimal, repo: "hexpm", optional: true]}], "hexpm", "c5eb0cab91f094599f94d55bc63409236a8ec69a21a67814529e8d5f6cc90b3b"},
+  "mime": {:hex, :mime, "2.0.7", "b8d739037be7cd402aee1ba0306edfdef982687ee7e9859bee6198c1e7e2f128", [:mix], [], "hexpm", "6171188e399ee16023ffc5b76ce445eb6d9672e2e241d2df6050f3c771e80ccd"},
+  "mint": {:hex, :mint, "1.7.1", "113fdb2b2f3b59e47c7955971854641c61f378549d73e829e1768de90fc1abf1", [:mix], [{:castore, "~> 0.1.0 or ~> 1.0", [hex: :castore, repo: "hexpm", optional: true]}, {:hpax, "~> 0.1.1 or ~> 0.2.0 or ~> 1.0", [hex: :hpax, repo: "hexpm", optional: false]}], "hexpm", "fceba0a4d0f24301ddee3024ae116df1c3f4bb7a563a731f45fdfeb9d39a231b"},
+  "nimble_options": {:hex, :nimble_options, "1.1.1", "e3a492d54d85fc3fd7c5baf411d9d2852922f66e69476317787a7b2bb000a61b", [:mix], [], "hexpm", "821b2470ca9442c4b6984882fe9bb0389371b8ddec4d45a9504f00a66f650b44"},
+  "nimble_pool": {:hex, :nimble_pool, "1.1.0", "bf9c29fbdcba3564a8b800d1eeb5a3c58f36e1e11d7b7fb2e084a643f645f06b", [:mix], [], "hexpm", "af2e4e6b34197db81f7aad230c1118eac993acc0dae6bc83bac0126d4ae0813a"},
+  "req": {:hex, :req, "0.5.17", "0096ddd5b0ed6f576a03dde4b158a0c727215b15d2795e59e0916c6971066ede", [:mix], [{:brotli, "~> 0.3.1", [hex: :brotli, repo: "hexpm", optional: true]}, {:ezstd, "~> 1.0", [hex: :ezstd, repo: "hexpm", optional: true]}, {:finch, "~> 0.17", [hex: :finch, repo: "hexpm", optional: false]}, {:jason, "~> 1.0", [hex: :jason, repo: "hexpm", optional: false]}, {:mime, "~> 2.0.6 or ~> 2.1", [hex: :mime, repo: "hexpm", optional: false]}, {:nimble_csv, "~> 1.0", [hex: :nimble_csv, repo: "hexpm", optional: true]}, {:plug, "~> 1.0", [hex: :plug, repo: "hexpm", optional: true]}], "hexpm", "0b8bc6ffdfebbc07968e59d3ff96d52f2202d0536f10fef4dc11dc02a2a43e39"},
+  "telemetry": {:hex, :telemetry, "1.3.0", "fedebbae410d715cf8e7062c96a1ef32ec22e764197f70cda73d82778d61e7a2", [:rebar3], [], "hexpm", "7015fc8919dbe63764f4b4b87a95b7c0996bd539e0d499be6ec9d7f3875b79e6"},
+}