Feat: Add basic export flow for Obsidian.

README.md

@@ -29,18 +29,20 @@ It runs a **4-stage AI pipeline** on your bookmarks:
     ↓
 🏷️  Entity Extraction   — mines hashtags, URLs, mentions, and 100+ known tools from raw tweet data (free, zero API calls)
     ↓
-👁️  Vision Analysis      — reads text, objects, and context from every image/GIF/video thumbnail (30–40 visual tags per image)
+👁️  Vision Analysis      — reads text, objects, and context from every image/GIF/video thumbnail
     ↓
-🧠 Semantic Tagging     — generates 25–35 searchable tags per bookmark for AI-powered search
+🧠 Semantic Tagging     — generates broad, reusable search tags per bookmark for AI-powered search
     ↓
 📂 Categorization       — assigns each bookmark to 1–3 categories with confidence scores
+    ↓
+📖 Obsidian Export      — writes your knowledge base as Markdown notes with YAML frontmatter, wikilinks, and index files
 ```
 
 After the pipeline runs, you get:
 - **AI search** — find bookmarks by meaning, not just keywords (*"funny meme about crypto crashing"*)
 - **Interactive mindmap** — explore your entire bookmark graph visually
 - **Filtered browsing** — grid or list view, filter by category, media type, and date
-- **Export tools** — download media, export as CSV / JSON / ZIP
+- **Export tools** — download media, export as CSV / JSON / ZIP / Obsidian vault
 
 ---
 
@@ -224,6 +226,45 @@ Create custom categories with a name, color, and optional description. The descr
 - **CSV** — spreadsheet-compatible with all fields
 - **JSON** — full structured data export
 - **ZIP** — exports a category's bookmarks + all media files with a `manifest.csv`
+- **Obsidian** — exports your entire knowledge base as a Markdown vault (see below)
+
+### 📖 Obsidian Export
+
+Export all processed bookmarks directly into an [Obsidian](https://obsidian.md) vault as structured Markdown notes.
+
+**Each bookmark becomes a note** with YAML frontmatter:
+
+```yaml
+---
+tweet_id: "1933508347334177246"
+author: "alex_prompter"
+author_name: "Alex Prompter"
+date: 2025-06-13
+source: "https://x.com/alex_prompter/status/1933508347334177246"
+categories: ["AI & Agents", "PKM & Workflows"]
+tags:
+  - twitter/bookmark
+  - author/alex_prompter
+  - prompt-engineering
+  - llm
+  - automation
+  - category/ai-agents
+---
+```
+
+**Index notes** are generated automatically in a `_index/` subfolder:
+- One note per **category** — lists all bookmarks in that category as `[[wikilinks]]`
+- One note per **author** — lists every bookmark from that person
+
+This creates a dense backlink graph in Obsidian's graph view, where bookmarks cluster naturally by topic and author.
+
+**To set up:**
+1. Go to **Settings** → find the **Obsidian Export** section
+2. Enter the full path to your Obsidian vault folder (e.g. `/Users/you/Obsidian/Personal`)
+3. Click **Save**, then **Export to Obsidian**
+4. Open your vault in Obsidian — notes appear in a `Twitter Bookmarks/` folder
+
+Re-export anytime. By default, existing notes are skipped — enable **Overwrite** to replace them.
 
 ### ⌨️ Command Palette
 
@@ -241,6 +282,7 @@ All settings are manageable in the **Settings** page at `/settings` or via envir
 | API Base URL | `ANTHROPIC_BASE_URL` | Custom endpoint for proxies or local Anthropic-compatible models |
 | AI Model | Settings page only | Haiku 4.5 (default, fastest/cheapest), Sonnet 4.6, Opus 4.6 |
 | OpenAI Key | Settings page only | Alternative provider if no Anthropic key is set |
+| Obsidian Vault Path | Settings page only | Absolute path to your Obsidian vault folder for Markdown export |
 | Database | `DATABASE_URL` | SQLite file path (default: `file:./prisma/dev.db`) |
 
 ### Custom API Endpoint
@@ -266,6 +308,7 @@ siftly/
 │   │   │   └── [slug]/       # Individual category operations
 │   │   ├── categorize/       # 4-stage AI pipeline (start, status, stop)
 │   │   ├── export/           # CSV, JSON, ZIP export
+│   │   │   └── obsidian/     # Obsidian vault export endpoint
 │   │   ├── import/           # JSON file import with dedup + auto-pipeline trigger
 │   │   │   ├── bookmarklet/  # Bookmarklet-specific import endpoint
 │   │   │   └── twitter/      # Twitter-specific import endpoint
@@ -308,6 +351,7 @@ siftly/
 │   ├── rawjson-extractor.ts  # Entity extraction from raw tweet JSON
 │   ├── parser.ts             # Multi-format JSON parser
 │   ├── exporter.ts           # CSV, JSON, ZIP export
+│   ├── obsidian-exporter.ts  # Obsidian vault export (Markdown + YAML frontmatter + indexes)
 │   ├── types.ts              # Shared TypeScript types
 │   └── db.ts                 # Prisma client singleton
 │

app/api/export/obsidian/route.ts

@@ -0,0 +1,38 @@
+import { NextRequest, NextResponse } from 'next/server'
+import { exportToObsidian } from '@/lib/obsidian-exporter'
+import prisma from '@/lib/db'
+
+export async function POST(request: NextRequest): Promise<NextResponse> {
+  let body: { category?: string; overwrite?: boolean } = {}
+  try {
+    body = await request.json()
+  } catch {
+    // body stays as defaults
+  }
+
+  const { category, overwrite = false } = body
+
+  const setting = await prisma.setting.findUnique({ where: { key: 'obsidianVaultPath' } })
+  if (!setting?.value) {
+    return NextResponse.json(
+      { error: 'Obsidian vault path not configured. Add it in Settings.' },
+      { status: 400 }
+    )
+  }
+
+  try {
+    const result = await exportToObsidian({
+      vaultPath: setting.value,
+      subfolder: 'Twitter Bookmarks',
+      overwrite,
+      categoryFilter: category,
+    })
+    return NextResponse.json(result)
+  } catch (err: unknown) {
+    console.error('Obsidian export error:', err)
+    return NextResponse.json(
+      { error: `Export failed: ${err instanceof Error ? err.message : String(err)}` },
+      { status: 500 }
+    )
+  }
+}

app/api/settings/route.ts

@@ -24,14 +24,15 @@ const ALLOWED_OPENAI_MODELS = [
 
 export async function GET(): Promise<NextResponse> {
   try {
-    const [anthropic, anthropicModel, provider, openai, openaiModel, xClientId, xClientSecret] = await Promise.all([
+    const [anthropic, anthropicModel, provider, openai, openaiModel, xClientId, xClientSecret, obsidianVaultPath] = await Promise.all([
       prisma.setting.findUnique({ where: { key: 'anthropicApiKey' } }),
       prisma.setting.findUnique({ where: { key: 'anthropicModel' } }),
       prisma.setting.findUnique({ where: { key: 'aiProvider' } }),
       prisma.setting.findUnique({ where: { key: 'openaiApiKey' } }),
       prisma.setting.findUnique({ where: { key: 'openaiModel' } }),
       prisma.setting.findUnique({ where: { key: 'x_oauth_client_id' } }),
       prisma.setting.findUnique({ where: { key: 'x_oauth_client_secret' } }),
+      prisma.setting.findUnique({ where: { key: 'obsidianVaultPath' } }),
     ])
 
     return NextResponse.json({
@@ -45,6 +46,7 @@ export async function GET(): Promise<NextResponse> {
       xOAuthClientId: maskKey(xClientId?.value ?? null),
       xOAuthClientSecret: maskKey(xClientSecret?.value ?? null),
       hasXOAuth: !!xClientId?.value,
+      obsidianVaultPath: obsidianVaultPath?.value ?? null,
     })
   } catch (err) {
     console.error('Settings GET error:', err)
@@ -161,6 +163,20 @@ export async function POST(request: NextRequest): Promise<NextResponse> {
     }
   }
 
+  // Save Obsidian vault path if provided
+  if ('obsidianVaultPath' in body) {
+    const vaultPath = (body as { obsidianVaultPath?: string }).obsidianVaultPath
+    if (typeof vaultPath !== 'string' || vaultPath.trim() === '') {
+      return NextResponse.json({ error: 'Invalid obsidianVaultPath value' }, { status: 400 })
+    }
+    await prisma.setting.upsert({
+      where: { key: 'obsidianVaultPath' },
+      update: { value: vaultPath.trim() },
+      create: { key: 'obsidianVaultPath', value: vaultPath.trim() },
+    })
+    return NextResponse.json({ saved: true })
+  }
+
   // Save X OAuth credentials if provided
   const { xOAuthClientId, xOAuthClientSecret } = body
   const xKeys: { key: string; value: string | undefined }[] = [

app/api/settings/test/route.ts

@@ -53,7 +53,7 @@ export async function POST(request: NextRequest): Promise<NextResponse> {
 
     let client
     try {
-      client = resolveOpenAIClient({ dbKey })
+      client = await resolveOpenAIClient({ dbKey })
     } catch {
       return NextResponse.json({ working: false, error: 'No OpenAI API key found. Add one in Settings or set up Codex CLI.' })
     }

app/import/page.tsx

@@ -92,7 +92,16 @@ const BOOKMARKLET_SCRIPT = `(async function(){
   function addTweet(t){
     if(!t||!t.rest_id||seen.has(t.rest_id))return;
     seen.add(t.rest_id);
-    var leg=t.legacy||{},usr=(t.core&&t.core.user_results&&t.core.user_results.result&&t.core.user_results.result.legacy)||{};
+    var leg=t.legacy||{};
+    var res=t.core&&t.core.user_results&&t.core.user_results.result;
+    var usrLeg=(res&&res.legacy)||{};
+    var usrCore=(res&&res.core)||{};
+    var usrAvatar=(res&&res.avatar)||{};
+    var usr={
+      name:usrCore.name||usrLeg.name||'Unknown',
+      screen_name:usrCore.screen_name||usrLeg.screen_name||'unknown',
+      profile_image_url_https:usrAvatar.image_url||usrLeg.profile_image_url_https||''
+    };
     var rawMedia=(leg.extended_entities&&leg.extended_entities.media)||(leg.entities&&leg.entities.media)||[];
     var media=rawMedia.map(function(m){
       var thumb=m.media_url_https||'';
@@ -191,19 +200,20 @@ const BOOKMARKLET_SCRIPT = `(async function(){
   document.body.appendChild(btn);
   document.body.appendChild(autoBtn);
   var origFetch=window.fetch;
+  function isRelevantUrl(u){return u.includes('/graphql/')&&(isLikes?u.toLowerCase().includes('like'):u.includes('Bookmark'));}
   window.fetch=async function(){
     var r=await origFetch.apply(this,arguments);
     try{
       var u=arguments[0] instanceof Request?arguments[0].url:String(arguments[0]);
-      if(u.includes('/graphql/')){var d=await r.clone().json();processData(d);}
+      if(isRelevantUrl(u)){var d=await r.clone().json();processData(d);}
     }catch(ex){}
     return r;
   };
   var origOpen=XMLHttpRequest.prototype.open,origSend=XMLHttpRequest.prototype.send,xhrUrls=new WeakMap();
   XMLHttpRequest.prototype.open=function(){xhrUrls.set(this,String(arguments[1]||''));return origOpen.apply(this,arguments);};
   XMLHttpRequest.prototype.send=function(){
     var xhr=this,u=xhrUrls.get(xhr)||'';
-    if(u.includes('/graphql/')){xhr.addEventListener('load',function(){try{processData(JSON.parse(xhr.responseText));}catch(ex){}});}
+    if(isRelevantUrl(u)){xhr.addEventListener('load',function(){try{processData(JSON.parse(xhr.responseText));}catch(ex){}});}
     return origSend.apply(this,arguments);
   };
   showToast('\u2705 Active! Scroll your '+label+' \u2014 counter updates above.','#1e1b4b');
@@ -222,7 +232,16 @@ const CONSOLE_SCRIPT = `(async function() {
   function addTweet(t) {
     if (!t?.rest_id || seen.has(t.rest_id)) return;
     seen.add(t.rest_id);
-    const leg = t.legacy ?? {}, usr = t.core?.user_results?.result?.legacy ?? {};
+    const leg = t.legacy ?? {};
+    const res = t.core?.user_results?.result;
+    const usrLeg = res?.legacy ?? {};
+    const usrCore = res?.core ?? {};
+    const usrAvatar = res?.avatar ?? {};
+    const usr = {
+      name: usrCore.name ?? usrLeg.name ?? 'Unknown',
+      screen_name: usrCore.screen_name ?? usrLeg.screen_name ?? 'unknown',
+      profile_image_url_https: usrAvatar.image_url ?? usrLeg.profile_image_url_https ?? '',
+    };
     const media = (leg.extended_entities?.media ?? leg.entities?.media ?? []).map(m => {
       const thumb = m.media_url_https ?? '';
       if (m.type === 'video' || m.type === 'animated_gif') {
@@ -336,14 +355,13 @@ const CONSOLE_SCRIPT = `(async function() {
     autoBtn.style.background = '#4f46e5'; autoBtn.style.color = '#fff'; autoBtn.style.border = 'none';
     runAutoScroll();
   };
-  document.body.appendChild(btn);
-  document.body.appendChild(autoBtn);
+  const isRelevantUrl = (u) => u.includes('/graphql/') && (isLikes ? u.toLowerCase().includes('like') : u.includes('Bookmark'));
   const origFetch = window.fetch;
   window.fetch = async function(...args) {
     const r = await origFetch.apply(this, args);
     try {
       const u = args[0] instanceof Request ? args[0].url : String(args[0]);
-      if (u.includes('/graphql/')) {
+      if (isRelevantUrl(u)) {
         const d = await r.clone().json();
         processData(d);
       }
@@ -359,13 +377,15 @@ const CONSOLE_SCRIPT = `(async function() {
   };
   XMLHttpRequest.prototype.send = function(...args) {
     const xhr = this, u = xhrUrls.get(xhr) ?? '';
-    if (u.includes('/graphql/')) {
+    if (isRelevantUrl(u)) {
       xhr.addEventListener('load', function() {
         try { processData(JSON.parse(xhr.responseText)); } catch(e) {}
       });
     }
     return origSend.apply(this, args);
   };
+  document.body.appendChild(btn);
+  document.body.appendChild(autoBtn);
   console.log(\`✅ Script active. Scroll through your \${label}, then click the purple button.\`);
 })();`