chore(deps): update dependency aiosqlite to v0.22.1

Reviewed-on: #1

.gitea/workflows/build.yml

@@ -27,6 +27,9 @@ jobs:
       - name: Log in to registry
         run: echo "${{ secrets.CI_TOKEN }}" | docker login gitea.gitea.svc.cluster.local:3000 -u chemavx --password-stdin
 
+      - name: Clean previous buildx builder
+        run: docker buildx rm ci-builder 2>/dev/null || true
+
       - name: Create buildx builder
         run: |
           cat > /tmp/buildkitd.toml << 'EOF'

Dockerfile

@@ -6,6 +6,7 @@ WORKDIR /app
 RUN apt-get update && apt-get install -y \
     gcc g++ \
     libxml2-dev libxslt-dev \
+    libfreetype6-dev \
     && rm -rf /var/lib/apt/lists/*
 
 COPY requirements.txt .

renovate.json

@@ -0,0 +1,3 @@
+{
+  "$schema": "https://docs.renovatebot.com/renovate-schema.json"
+}

requirements.txt

@@ -8,15 +8,15 @@ aiohttp==3.10.0
 # Scraping
 beautifulsoup4==4.12.3
 lxml==5.2.2
-trafilatura==1.12.0
+trafilatura==1.12.2
 youtube-transcript-api==0.6.2
-pdfplumber==0.11.3
+pdfplumber==0.11.9
-feedparser==6.0.11
+feedparser==6.0.12
 duckduckgo-search==6.2.6
 
 # Storage & Embeddings
-sqlite-vec==0.1.6
+sqlite-vec==0.1.9
-aiosqlite==0.20.0
+aiosqlite==0.22.1
 
 # Processing
 tiktoken==0.7.0
@@ -26,6 +26,10 @@ scikit-learn==1.5.1
 # Claude API (scoring)
 anthropic>=0.40.0
 
+# PDF export
+markdown==3.7
+reportlab==4.2.5
+
 # Utilities
 pydantic==2.8.0
 pydantic-settings==2.4.0

src/bot/bot.py

@@ -4,6 +4,7 @@ Main user interface — all commands handled here
 """
 import asyncio
 import os
+import time
 from datetime import datetime
 from typing import Optional
 
@@ -33,17 +34,29 @@ def is_authorized(user_id: int) -> bool:
     return not allowed or user_id in allowed
 
 
-def fmt_progress(iteration: int, total: int, new: int, stats: dict) -> str:
+class ProgressReporter:
-    scraped = stats.get("scraped") or 0
+    def __init__(self, reply_target: Message = None, *, bot=None, chat_id: int = None):
-    failed  = stats.get("failed")  or 0
+        self._reply_target = reply_target
-    pending = stats.get("pending") or 0
+        self._bot = bot
-    skipped = stats.get("skipped") or 0
+        self._chat_id = chat_id
-    return (
+        self._msg: Optional[Message] = None
-        f"🔄 *Iteration {iteration}*\n"
+
-        f"📚 Sources found: `{total}`\n"
+    async def start(self, text: str):
-        f"✅ Scraped: `{scraped}` | ⏭️ Skipped: `{skipped}` | ❌ Failed: `{failed}` | ⏳ Pending: `{pending}`\n"
+        if self._reply_target is not None:
-        f"🆕 New URLs this round: `{new}`"
+            self._msg = await self._reply_target.reply_text(text, parse_mode=ParseMode.MARKDOWN)
-    )
+        elif self._bot is not None and self._chat_id is not None:
+            self._msg = await self._bot.send_message(self._chat_id, text, parse_mode=ParseMode.MARKDOWN)
+
+    async def update(self, text: str):
+        if not self._msg:
+            return
+        try:
+            await self._msg.edit_text(text, parse_mode=ParseMode.MARKDOWN)
+        except Exception:
+            pass
+
+    async def done(self, text: str):
+        await self.update(text)
 
 
 async def send_chunked(message: Message, text: str, parse_mode=None):
@@ -56,6 +69,72 @@ async def send_chunked(message: Message, text: str, parse_mode=None):
             await asyncio.sleep(0.5)
 
 
+# ─── Shared research logic ────────────────────────────────────────────────────
+
+async def run_scheduled_research(bot, chat_id: int, topic: str,
+                                  session_id: int, db: ResearchDB,
+                                  progress_message=None,
+                                  silent_completion: bool = False):
+    if progress_message is not None:
+        reporter = ProgressReporter(progress_message)
+    else:
+        reporter = ProgressReporter(bot=bot, chat_id=chat_id)
+
+    try:
+        await reporter.start(f"🔍 Iniciando scraping de `{topic}`…")
+
+        async def on_progress(iter_num, total_sources):
+            await reporter.update(
+                f"🔍 Scraping — iteración `{iter_num}` | `{total_sources}` fuentes encontradas"
+            )
+
+        scraper = ExhaustiveScraper(db, session_id, topic, on_progress)
+        final_stats = await scraper.run()
+
+        await db.update_session(session_id, status=ResearchStatus.SATURATED)
+        scraped = final_stats.get("scraped", 0)
+
+        await reporter.update(f"⚡ Procesando `{scraped}` fuentes…")
+
+        ollama = OllamaClient()
+        if await ollama.is_available():
+            processor = ContentProcessor(db, ollama)
+
+            async def proc_progress(total_chunks, total_words):
+                await reporter.update(
+                    f"⚡ Scoring chunks… (`{total_chunks}` procesados)"
+                )
+
+            await processor.process_session(session_id, topic, proc_progress)
+            chunk_count = await db.get_chunks_count(session_id)
+            if silent_completion:
+                await reporter.done(
+                    f"🔍 Investigación completada — analizando novedades…"
+                )
+            else:
+                await reporter.done(
+                    f"✅ Listo — `{scraped}` fuentes · `{chunk_count}` chunks · usa /generate <tipo>"
+                )
+        else:
+            await reporter.done(
+                f"⚠️ Ollama no disponible — `{scraped}` fuentes scraped.\n"
+                f"Usa /generate para generar contenido."
+            )
+
+    except asyncio.CancelledError:
+        await db.update_session(session_id, status=ResearchStatus.FINISHED)
+        try:
+            await reporter.done("🛑 Investigación cancelada.")
+        except Exception:
+            pass
+    except Exception as e:
+        logger.error("Research task failed", error=str(e))
+        try:
+            await reporter.done(f"❌ Error: {str(e)[:200]}")
+        except Exception:
+            pass
+
+
 # ─── Commands ─────────────────────────────────────────────────────────────────
 
 async def cmd_start(update: Update, ctx: ContextTypes.DEFAULT_TYPE):
@@ -68,9 +147,18 @@ async def cmd_start(update: Update, ctx: ContextTypes.DEFAULT_TYPE):
         "`/status` — Check current research progress\n"
         "`/finish` — Stop research and proceed to generation\n"
         "`/process` — Manually trigger chunk processing\n"
-        "`/generate <type>` — Generate output (podcast|blog|report|thread)\n"
+        "`/generate <type>` — Generate output\n"
+        "  Tipos: podcast|blog|report|thread\n"
+        "  Extended: podcast_extended|blog_extended|report_extended\n"
         "`/sources` — List all sources found\n"
         "`/outputs` — List generated outputs\n"
+        "`/export` — Exportar último output como PDF\n"
+        "`/publish` — Publicar último output en Ghost como borrador\n"
+        "`/compare <tema1> vs <tema2>` — Análisis comparativo\n"
+        "`/costs` — Show API usage costs\n"
+        "`/watch <topic> [h]` — Schedule periodic research\n"
+        "`/unwatch <topic>` — Remove a watch\n"
+        "`/watches` — List your watched topics\n"
         "`/cancel` — Cancel current research\n"
         "`/help` — Show this message",
         parse_mode=ParseMode.MARKDOWN
@@ -91,88 +179,22 @@ async def cmd_research(update: Update, ctx: ContextTypes.DEFAULT_TYPE):
         )
         return
 
-    # Check for existing active research
     if chat_id in _active_tasks and not _active_tasks[chat_id].done():
         await update.message.reply_text(
             "⚠️ Research already in progress. Use /status or /finish first."
         )
         return
 
-    msg = await update.message.reply_text(
-        f"🦉 *ResearchOwl* starting research on:\n`{topic}`\n\n"
-        f"🌱 Seeding sources from:\n"
-        f"• DuckDuckGo (8 queries)\n"
-        f"• Wikipedia + internal links\n"
-        f"• Reddit top posts\n"
-        f"• YouTube transcripts\n\n"
-        f"This will run exhaustively until saturation. Use /finish to stop early.",
-        parse_mode=ParseMode.MARKDOWN
-    )
-
     async def run_research():
         db_conn = await get_db()
         db = ResearchDB(db_conn)
         try:
             session_id = await db.create_session(topic, chat_id)
             _active_sessions[chat_id] = session_id
-
+            await run_scheduled_research(
-            progress_msg = msg
+                ctx.bot, chat_id, topic, session_id, db,
-            iteration_count = [0]
+                progress_message=update.message
-
-            async def on_progress(iteration, total, new_this_round, stats):
-                iteration_count[0] = iteration
-                text = fmt_progress(iteration, total, new_this_round, stats)
-                try:
-                    await progress_msg.edit_text(text, parse_mode=ParseMode.MARKDOWN)
-                except Exception:
-                    pass
-
-            scraper = ExhaustiveScraper(db, session_id, topic, on_progress)
-            final_stats = await scraper.run()
-
-            await db.update_session(session_id, status=ResearchStatus.SATURATED)
-
-            scraped = final_stats.get("scraped", 0)
-            await update.message.reply_text(
-                f"✅ *Research complete!*\n\n"
-                f"📊 Results:\n"
-                f"• Sources found & scraped: `{scraped}`\n"
-                f"• Iterations: `{iteration_count[0]}`\n\n"
-                f"Now processing content with Ollama...\n"
-                f"Use `/generate podcast|blog|report|thread` when ready.",
-                parse_mode=ParseMode.MARKDOWN
             )
-
-            # Auto-process after scraping
-            ollama = OllamaClient()
-            if await ollama.is_available():
-                processor = ContentProcessor(db, ollama)
-
-                async def proc_progress(total_chunks, total_words):
-                    await update.message.reply_text(
-                        f"🧠 *Processing complete!*\n"
-                        f"• Chunks stored: `{total_chunks}`\n"
-                        f"• Words researched: `{total_words:,}`\n\n"
-                        f"Ready! Use `/generate podcast|blog|report|thread`",
-                        parse_mode=ParseMode.MARKDOWN
-                    )
-
-                await processor.process_session(session_id, topic, proc_progress)
-            else:
-                await update.message.reply_text(
-                    "⚠️ Ollama not reachable — skipping processing.\n"
-                    "You can still use `/generate` (will use raw content)."
-                )
-
-        except asyncio.CancelledError:
-            await db.update_session(
-                _active_sessions.get(chat_id, 0),
-                status=ResearchStatus.FINISHED
-            )
-            await update.message.reply_text("🛑 Research cancelled.")
-        except Exception as e:
-            logger.error("Research task failed", error=str(e))
-            await update.message.reply_text(f"❌ Research failed: {str(e)[:200]}")
         finally:
             await db_conn.close()
 
@@ -255,6 +277,7 @@ async def cmd_generate(update: Update, ctx: ContextTypes.DEFAULT_TYPE):
 
     chat_id = update.effective_chat.id
     output_arg = ctx.args[0].lower() if ctx.args else ""
+    lang = "en" if len(ctx.args) > 1 and ctx.args[1].lower() == "en" else "es"
 
     type_map = {
         "podcast": OutputType.PODCAST,
@@ -263,6 +286,10 @@ async def cmd_generate(update: Update, ctx: ContextTypes.DEFAULT_TYPE):
         "thread": OutputType.THREAD,
         "hilo": OutputType.THREAD,
         "informe": OutputType.REPORT,
+        "report_extended":  OutputType.REPORT_EXTENDED,
+        "blog_extended":    OutputType.BLOG_EXTENDED,
+        "podcast_extended": OutputType.PODCAST_EXTENDED,
+        "informe_extended": OutputType.REPORT_EXTENDED,
     }
 
     if output_arg not in type_map:
@@ -279,7 +306,14 @@ async def cmd_generate(update: Update, ctx: ContextTypes.DEFAULT_TYPE):
     db = ResearchDB(db_conn)
 
     try:
-        # Find last session for this chat
+        # Usa la sesión activa si existe, si no la más reciente
+        session_id = _active_sessions.get(chat_id)
+        if session_id:
+            cursor = await db_conn.execute(
+                "SELECT * FROM research_sessions WHERE id = ?",
+                (session_id,)
+            )
+        else:
             cursor = await db_conn.execute(
                 """SELECT * FROM research_sessions WHERE telegram_chat_id = ?
                    ORDER BY created_at DESC LIMIT 1""",
@@ -293,9 +327,11 @@ async def cmd_generate(update: Update, ctx: ContextTypes.DEFAULT_TYPE):
         session = dict(row)
         session_id = session["id"]
 
+        backend = "Claude Haiku" if settings.anthropic_api_key else f"Ollama ({settings.ollama_model})"
+        lang_label = " (EN)" if lang == "en" else ""
         msg = await update.message.reply_text(
-            f"⚙️ Generating *{output_type}* for: `{session['topic']}`\n"
+            f"⚙️ Generating *{output_type}{lang_label}* for: `{session['topic']}`\n"
-            f"Using Ollama ({settings.ollama_model})...\n"
+            f"Using {backend}...\n"
             f"This may take 2-5 minutes ☕",
             parse_mode=ParseMode.MARKDOWN
         )
@@ -310,18 +346,26 @@ async def cmd_generate(update: Update, ctx: ContextTypes.DEFAULT_TYPE):
         processor = ContentProcessor(db, ollama)
         generator = OutputGenerator(db, ollama, processor)
 
-        output = await generator.generate(session_id, output_type, gen_progress)
+        output = await generator.generate(session_id, output_type, gen_progress, lang=lang)
 
         # Send as file if very long
         if len(output) > 8000:
             import tempfile
+            import re as _re
             ext_map = {
                 OutputType.PODCAST: "script.md",
                 OutputType.BLOG: "post.md",
                 OutputType.REPORT: "report.md",
                 OutputType.THREAD: "thread.txt",
+                OutputType.REPORT_EXTENDED: "report_extended.md",
+                OutputType.BLOG_EXTENDED: "blog_extended.md",
+                OutputType.PODCAST_EXTENDED: "script_extended.md",
             }
-            filename = f"researchowl_{session['topic'][:30].replace(' ', '_')}_{ext_map[output_type]}"
+            # Use the topic from the output header (written at generation time)
+            # instead of the pre-fetched session dict which may be stale.
+            _m = _re.search(r'^Topic:\s*(.+)$', output[:500], _re.MULTILINE)
+            _topic = _m.group(1).strip() if _m else session["topic"]
+            filename = f"researchowl_{_topic[:30].replace(' ', '_')}_{ext_map[output_type]}"
 
             with tempfile.NamedTemporaryFile(mode="w", suffix=".md", delete=False) as f:
                 f.write(output)
@@ -339,6 +383,18 @@ async def cmd_generate(update: Update, ctx: ContextTypes.DEFAULT_TYPE):
         else:
             await send_chunked(update.message, output)
 
+        try:
+            stats = await db.get_usage_stats(session_id)
+            total_cost = sum(s.get("total_cost", 0) for s in stats)
+            if total_cost > settings.cost_alert_threshold:
+                await update.message.reply_text(
+                    f"⚠️ Coste acumulado de esta sesión: `${total_cost:.4f}`"
+                    f" (umbral: `${settings.cost_alert_threshold:.2f}`)",
+                    parse_mode=ParseMode.MARKDOWN
+                )
+        except Exception:
+            pass
+
     except Exception as e:
         logger.error("Generate failed", error=str(e))
         await update.message.reply_text(f"❌ Generation failed: {str(e)[:200]}")
@@ -429,6 +485,169 @@ async def cmd_outputs(update: Update, ctx: ContextTypes.DEFAULT_TYPE):
         await db_conn.close()
 
 
+async def cmd_costs(update: Update, ctx: ContextTypes.DEFAULT_TYPE):
+    if not is_authorized(update.effective_user.id):
+        return
+
+    chat_id = update.effective_chat.id
+    db_conn = await get_db()
+    db = ResearchDB(db_conn)
+
+    try:
+        cursor = await db_conn.execute(
+            "SELECT * FROM research_sessions WHERE telegram_chat_id = ? ORDER BY created_at DESC LIMIT 1",
+            (chat_id,)
+        )
+        row = await cursor.fetchone()
+        if not row:
+            await update.message.reply_text("No sessions found.")
+            return
+
+        session_id = row["id"]
+        topic = row["topic"]
+
+        by_type = {r["call_type"]: r for r in await db.get_usage_stats(session_id)}
+        totals = await db.get_total_usage_stats()
+
+        lines = [f"📊 *Costes ResearchOwl*\n"]
+        lines.append(f"Última sesión (`{topic}`):")
+
+        session_total = 0.0
+        for call_type, label in [("scoring", "Scoring"), ("generation", "Generación")]:
+            row_data = by_type.get(call_type)
+            if row_data:
+                calls = row_data["calls"]
+                tokens = row_data["total_tokens"]
+                cost = row_data["total_cost"]
+                session_total += cost
+                lines.append(f"  {label}:    {calls} llamadas · {tokens:,} tokens · ${cost:.4f}")
+            else:
+                lines.append(f"  {label}:    —")
+
+        lines.append(f"  Total:      ${session_total:.4f}")
+        lines.append("")
+        lines.append("Acumulado total:")
+        acc_cost = totals.get("total_cost") or 0.0
+        acc_sessions = totals.get("sessions") or 0
+        lines.append(f"  ${acc_cost:.4f} ({acc_sessions} sesiones)")
+
+        await update.message.reply_text("\n".join(lines), parse_mode=ParseMode.MARKDOWN)
+    finally:
+        await db_conn.close()
+
+
+async def cmd_watch(update: Update, ctx: ContextTypes.DEFAULT_TYPE):
+    if not is_authorized(update.effective_user.id):
+        return
+
+    chat_id = update.effective_chat.id
+    args = ctx.args or []
+
+    if not args:
+        await update.message.reply_text(
+            "❌ Uso: `/watch <tema> [horas]`\nEjemplo: `/watch Incidente Roswell 24`",
+            parse_mode=ParseMode.MARKDOWN
+        )
+        return
+
+    interval_hours = 24
+    if args[-1].isdigit():
+        interval_hours = int(args[-1])
+        topic = " ".join(args[:-1]).strip()
+    else:
+        topic = " ".join(args).strip()
+
+    if not topic:
+        await update.message.reply_text("❌ Debes especificar un tema.")
+        return
+
+    if not (1 <= interval_hours <= 168):
+        await update.message.reply_text(
+            "❌ El intervalo debe estar entre 1 y 168 horas (1 semana)."
+        )
+        return
+
+    db_conn = await get_db()
+    db = ResearchDB(db_conn)
+    try:
+        try:
+            await db.add_watch(topic, chat_id, interval_hours)
+            await update.message.reply_text(
+                f"👁 Watching: `{topic}` — cada {interval_hours}h\n"
+                f"Primera ejecución en ~{interval_hours}h.\n"
+                f"Usa /watches para ver todos tus temas.",
+                parse_mode=ParseMode.MARKDOWN
+            )
+        except Exception as e:
+            if "UNIQUE" in str(e):
+                await update.message.reply_text(
+                    f"Ya estás watching `{topic}`", parse_mode=ParseMode.MARKDOWN
+                )
+            else:
+                raise
+    finally:
+        await db_conn.close()
+
+
+async def cmd_unwatch(update: Update, ctx: ContextTypes.DEFAULT_TYPE):
+    if not is_authorized(update.effective_user.id):
+        return
+
+    chat_id = update.effective_chat.id
+    topic = " ".join(ctx.args).strip() if ctx.args else ""
+
+    if not topic:
+        await update.message.reply_text(
+            "❌ Uso: `/unwatch <tema>`", parse_mode=ParseMode.MARKDOWN
+        )
+        return
+
+    db_conn = await get_db()
+    db = ResearchDB(db_conn)
+    try:
+        removed = await db.remove_watch(topic, chat_id)
+        if removed:
+            await update.message.reply_text(
+                f"✅ Ya no vigilas `{topic}`.", parse_mode=ParseMode.MARKDOWN
+            )
+        else:
+            await update.message.reply_text(f"No estabas watching `{topic}`.")
+    finally:
+        await db_conn.close()
+
+
+async def cmd_watches(update: Update, ctx: ContextTypes.DEFAULT_TYPE):
+    if not is_authorized(update.effective_user.id):
+        return
+
+    chat_id = update.effective_chat.id
+    db_conn = await get_db()
+    db = ResearchDB(db_conn)
+    try:
+        watches = await db.list_watches(chat_id)
+        if not watches:
+            await update.message.reply_text(
+                "No tienes temas vigilados. Usa `/watch <tema>`",
+                parse_mode=ParseMode.MARKDOWN
+            )
+            return
+
+        now = time.time()
+        lines = ["👁 *Tus temas vigilados:*\n"]
+        for i, w in enumerate(watches, 1):
+            secs_remaining = max(0.0, w["next_run_at"] - now)
+            hours_remaining = secs_remaining / 3600
+            eta = f"{int(secs_remaining / 60)}min" if hours_remaining < 1 else f"{hours_remaining:.1f}h"
+            status = "✅" if w["enabled"] else "⏸"
+            lines.append(
+                f"{i}. {status} `{w['topic']}` — cada {w['interval_hours']}h · próxima en {eta}"
+            )
+
+        await update.message.reply_text("\n".join(lines), parse_mode=ParseMode.MARKDOWN)
+    finally:
+        await db_conn.close()
+
+
 async def cmd_process(update: Update, ctx: ContextTypes.DEFAULT_TYPE):
     if not is_authorized(update.effective_user.id):
         return
@@ -513,8 +732,448 @@ async def cmd_help(update: Update, ctx: ContextTypes.DEFAULT_TYPE):
 
 # ─── Bot setup ────────────────────────────────────────────────────────────────
 
+async def _purge_on_startup(app: Application) -> None:
+    db_conn = await get_db()
+    try:
+        db = ResearchDB(db_conn)
+        result = await db.purge_old_sessions(30)
+        if result["sessions"] > 0:
+            logger.info("Startup purge done", **result)
+    except Exception as e:
+        logger.warning("Startup purge failed — bot continues", error=str(e))
+    finally:
+        await db_conn.close()
+
+
+async def _scheduler_loop(app: Application):
+    while True:
+        db_conn = None
+        try:
+            db_conn = await get_db()
+            db = ResearchDB(db_conn)
+            due = await db.get_due_watches()
+            for watch in due:
+                chat_id = watch["chat_id"]
+                topic = watch["topic"]
+                if chat_id in _active_tasks and not _active_tasks[chat_id].done():
+                    continue
+                session_id = await db.create_session(topic, chat_id)
+                _active_sessions[chat_id] = session_id
+                await db.update_watch_run(watch["id"])
+
+                async def _task(c=chat_id, t=topic, s=session_id, w_id=watch["id"]):
+                    inner_db_conn = await get_db()
+                    inner_db = ResearchDB(inner_db_conn)
+                    try:
+                        await run_scheduled_research(app.bot, c, t, s, inner_db,
+                                                     silent_completion=True)
+
+                        prev_session = await inner_db.get_previous_session(c, t, s)
+                        new_urls = await inner_db.get_session_urls(s)
+                        old_urls = await inner_db.get_session_urls(prev_session["id"]) \
+                                   if prev_session else set()
+                        new_chunks = await inner_db.get_top_chunks(s, limit=30)
+
+                        try:
+                            from src.generator.generator import generate_diff_summary
+                            summary = await generate_diff_summary(
+                                t, new_urls, old_urls, new_chunks, s, inner_db
+                            )
+                        except Exception as e:
+                            logger.warning("Diff summary failed", error=str(e))
+                            summary = (
+                                f"📊 *Actualización disponible — {t}*\n\n"
+                                f"Usa /generate report para ver el análisis completo."
+                            )
+
+                        if summary:
+                            await app.bot.send_message(
+                                c, summary, parse_mode=ParseMode.MARKDOWN
+                            )
+                        else:
+                            await app.bot.send_message(
+                                c,
+                                f"🔄 *{t}* — sin novedades significativas esta vez.",
+                                parse_mode=ParseMode.MARKDOWN
+                            )
+                    finally:
+                        await inner_db_conn.close()
+
+                task = asyncio.create_task(_task())
+                _active_tasks[chat_id] = task
+                await app.bot.send_message(
+                    chat_id,
+                    f"🔄 Investigación automática iniciada: `{topic}`",
+                    parse_mode=ParseMode.MARKDOWN
+                )
+        except Exception as e:
+            logger.warning("Scheduler loop error", error=str(e))
+        finally:
+            if db_conn:
+                try:
+                    await db_conn.close()
+                except Exception:
+                    pass
+        await asyncio.sleep(60)
+
+
+async def _start_scheduler(app: Application) -> None:
+    asyncio.create_task(_scheduler_loop(app))
+
+
+async def _on_startup(app: Application) -> None:
+    await _purge_on_startup(app)
+    await _start_scheduler(app)
+
+
+async def cmd_export(update: Update, ctx: ContextTypes.DEFAULT_TYPE):
+    if not is_authorized(update.effective_user.id):
+        return
+
+    chat_id = update.effective_chat.id
+    db_conn = await get_db()
+    db = ResearchDB(db_conn)
+
+    try:
+        session = await db.get_latest_session(chat_id)
+        if not session:
+            await update.message.reply_text("No hay sesiones de investigación.")
+            return
+
+        session_id = session["id"]
+        topic = session["topic"]
+
+        outputs = await db.get_outputs(session_id)
+        if not outputs:
+            await update.message.reply_text(
+                "No hay outputs generados. Usa `/generate <tipo>` primero.",
+                parse_mode=ParseMode.MARKDOWN
+            )
+            return
+
+        priority = [
+            "report_extended", "blog_extended", "podcast_extended",
+            "report", "blog", "podcast", "thread",
+        ]
+        chosen = None
+        for ptype in priority:
+            for o in outputs:
+                if o["output_type"] == ptype:
+                    chosen = o
+                    break
+            if chosen:
+                break
+        if not chosen:
+            chosen = outputs[0]
+
+        msg = await update.message.reply_text(
+            f"📄 Generando PDF para `{topic}`…",
+            parse_mode=ParseMode.MARKDOWN
+        )
+
+        try:
+            from src.generator.generator import generate_pdf
+            pdf_bytes = generate_pdf(chosen["content"], title=topic)
+        except ImportError:
+            await msg.edit_text("❌ reportlab no está instalado. Ejecuta: `pip install reportlab`")
+            return
+        except Exception as e:
+            await msg.edit_text(f"❌ Error generando PDF: {str(e)[:200]}")
+            return
+
+        safe_topic = topic[:40].replace(" ", "_").replace("/", "-")
+        filename = f"researchowl_{safe_topic}_{chosen['output_type']}.pdf"
+
+        import io
+        await update.message.reply_document(
+            document=io.BytesIO(pdf_bytes),
+            filename=filename,
+            caption=f"📄 *{chosen['output_type'].upper()}* — {topic}\nExportado por ResearchOwl 🦉",
+            parse_mode=ParseMode.MARKDOWN
+        )
+        try:
+            await msg.delete()
+        except Exception:
+            pass
+
+    except Exception as e:
+        logger.error("Export failed", error=str(e))
+        await update.message.reply_text(f"❌ Export failed: {str(e)[:200]}")
+    finally:
+        await db_conn.close()
+
+
+async def cmd_purge(update: Update, ctx: ContextTypes.DEFAULT_TYPE):
+    if not is_authorized(update.effective_user.id):
+        return
+
+    args = ctx.args or []
+
+    if not args:
+        days = 30
+    else:
+        try:
+            days = int(args[0])
+        except ValueError:
+            await update.message.reply_text(
+                "❌ Uso: `/purge [días]`", parse_mode=ParseMode.MARKDOWN
+            )
+            return
+        if days < 0:
+            await update.message.reply_text("❌ El número de días debe ser ≥ 0.")
+            return
+        if days == 0 and not (len(args) >= 2 and args[1] == "confirm"):
+            await update.message.reply_text(
+                "⚠️ Esto borrará *todas* las sesiones completadas.\n"
+                "Envía `/purge 0 confirm` para confirmar.",
+                parse_mode=ParseMode.MARKDOWN
+            )
+            return
+
+    db_conn = await get_db()
+    try:
+        db = ResearchDB(db_conn)
+        result = await db.purge_old_sessions(days)
+        await update.message.reply_text(
+            f"🗑️ Purged: {result['sessions']} sessions, "
+            f"{result['sources']} sources, "
+            f"{result['chunks']} chunks, "
+            f"{result['outputs']} outputs"
+        )
+    except Exception as e:
+        logger.error("Purge command failed", error=str(e))
+        await update.message.reply_text(f"❌ Purge failed: {str(e)[:200]}")
+    finally:
+        await db_conn.close()
+
+
+async def cmd_publish(update: Update, ctx: ContextTypes.DEFAULT_TYPE):
+    if not is_authorized(update.effective_user.id):
+        return
+
+    chat_id = update.effective_chat.id
+    db_conn = await get_db()
+    db = ResearchDB(db_conn)
+
+    try:
+        from src.generator.generator import GhostPublisher, _extract_title
+
+        ghost = GhostPublisher()
+        if not ghost.is_configured():
+            await update.message.reply_text(
+                "❌ Ghost no configurado. Asegúrate de que `GHOST_URL` y `GHOST_API_KEY` están definidos.",
+                parse_mode=ParseMode.MARKDOWN
+            )
+            return
+
+        session_id = _active_sessions.get(chat_id)
+        if session_id:
+            cursor = await db_conn.execute(
+                "SELECT * FROM research_sessions WHERE id = ?",
+                (session_id,)
+            )
+        else:
+            cursor = await db_conn.execute(
+                "SELECT * FROM research_sessions WHERE telegram_chat_id = ? ORDER BY created_at DESC LIMIT 1",
+                (chat_id,)
+            )
+        row = await cursor.fetchone()
+        if not row:
+            await update.message.reply_text("No hay sesiones. Usa /research primero.")
+            return
+
+        session = dict(row)
+        outputs = await db.get_outputs(session["id"])
+        if not outputs:
+            await update.message.reply_text(
+                "No hay outputs generados. Usa `/generate blog|report` primero.",
+                parse_mode=ParseMode.MARKDOWN
+            )
+            return
+
+        priority = ["blog_extended", "blog", "report_extended", "report",
+                    "podcast_extended", "podcast", "thread"]
+        chosen = None
+        for ptype in priority:
+            for o in outputs:
+                if o["output_type"] == ptype:
+                    chosen = o
+                    break
+            if chosen:
+                break
+        if not chosen:
+            chosen = outputs[-1]
+
+        msg = await update.message.reply_text("📤 Publicando en Ghost como borrador…")
+
+        title = _extract_title(chosen["content"]) or session["topic"]
+        result = await ghost.publish_draft(title, chosen["content"])
+        post = result["posts"][0]
+        admin_url = f"{ghost.url}/ghost/#/editor/post/{post['id']}"
+
+        await msg.edit_text(
+            f"✅ *Publicado en Ghost como borrador*\n\n"
+            f"📝 Título: `{title}`\n"
+            f"🔗 Editar: {admin_url}",
+            parse_mode=ParseMode.MARKDOWN
+        )
+
+    except Exception as e:
+        logger.error("Publish to Ghost failed", error=str(e))
+        await update.message.reply_text(f"❌ Error publicando en Ghost: {str(e)[:200]}")
+    finally:
+        await db_conn.close()
+
+
+async def cmd_compare(update: Update, ctx: ContextTypes.DEFAULT_TYPE):
+    if not is_authorized(update.effective_user.id):
+        return
+
+    chat_id = update.effective_chat.id
+    text = " ".join(ctx.args).strip() if ctx.args else ""
+
+    import re
+    match = re.split(r'\s+vs\.?\s+|\s+versus\s+', text, maxsplit=1, flags=re.IGNORECASE)
+    if len(match) != 2 or not match[0].strip() or not match[1].strip():
+        await update.message.reply_text(
+            "❌ Uso: `/compare <tema1> vs <tema2>`\n"
+            "Ejemplo: `/compare energía solar vs energía nuclear`",
+            parse_mode=ParseMode.MARKDOWN
+        )
+        return
+
+    topic_a = match[0].strip()
+    topic_b = match[1].strip()
+
+    if chat_id in _active_tasks and not _active_tasks[chat_id].done():
+        await update.message.reply_text(
+            "⚠️ Ya hay una investigación en curso. Usa /cancel primero."
+        )
+        return
+
+    msg = await update.message.reply_text(
+        f"🔍 Comparando `{topic_a}` vs `{topic_b}`…\n"
+        f"Esto lanzará dos investigaciones en paralelo y tardará varios minutos.",
+        parse_mode=ParseMode.MARKDOWN
+    )
+
+    async def run_compare():
+        db_conn_a = await get_db()
+        db_conn_b = await get_db()
+        db_a = ResearchDB(db_conn_a)
+        db_b = ResearchDB(db_conn_b)
+
+        try:
+            session_id_a = await db_a.create_session(topic_a, chat_id)
+            session_id_b = await db_b.create_session(topic_b, chat_id)
+            _active_sessions[chat_id] = session_id_a
+
+            await msg.edit_text(
+                f"🔍 Investigando en paralelo:\n"
+                f"• `{topic_a}`\n"
+                f"• `{topic_b}`\n\n"
+                f"Esto puede tardar 10-20 minutos…",
+                parse_mode=ParseMode.MARKDOWN
+            )
+
+            async def research_topic(session_id, topic, db):
+                scraper = ExhaustiveScraper(db, session_id, topic)
+                await scraper.run()
+                await db.update_session(session_id, status=ResearchStatus.SATURATED)
+                ollama = OllamaClient()
+                if await ollama.is_available():
+                    processor = ContentProcessor(db, ollama)
+                    await processor.process_session(session_id, topic)
+
+            await msg.edit_text(
+                f"🔍 Scraping en paralelo:\n• `{topic_a}`\n• `{topic_b}`…",
+                parse_mode=ParseMode.MARKDOWN
+            )
+
+            await asyncio.gather(
+                research_topic(session_id_a, topic_a, db_a),
+                research_topic(session_id_b, topic_b, db_b),
+            )
+
+            await msg.edit_text(
+                "✍️ Generando análisis comparativo…",
+                parse_mode=ParseMode.MARKDOWN
+            )
+
+            ollama = OllamaClient()
+            processor_a = ContentProcessor(db_a, ollama)
+            processor_b = ContentProcessor(db_b, ollama)
+
+            context_a = await processor_a.rag_query(session_id_a, topic_a, top_k=40)
+            context_b = await processor_b.rag_query(session_id_b, topic_b, top_k=40)
+
+            if not context_a:
+                chunks = await db_a.get_top_chunks(session_id_a, limit=20)
+                context_a = "\n\n---\n\n".join(c["content"] for c in chunks)
+            if not context_b:
+                chunks = await db_b.get_top_chunks(session_id_b, limit=20)
+                context_b = "\n\n---\n\n".join(c["content"] for c in chunks)
+
+            from src.generator.generator import generate_comparison
+            comparison = await generate_comparison(
+                topic_a, topic_b, context_a, context_b, session_id_a, db_a
+            )
+
+            from datetime import datetime, timezone
+            now = datetime.now(timezone.utc).strftime("%Y-%m-%d %H:%M UTC")
+            header = (
+                f"---\n"
+                f"ResearchOwl | COMPARISON\n"
+                f"Topic A: {topic_a}\n"
+                f"Topic B: {topic_b}\n"
+                f"Generated: {now}\n"
+                f"---\n\n"
+            )
+            full_output = header + comparison
+
+            await db_a.save_output(session_id_a, OutputType.REPORT, full_output)
+
+            if len(full_output) > 8000:
+                import io
+                filename = (
+                    f"compare_{topic_a[:20]}_{topic_b[:20]}.md"
+                    .replace(" ", "_")
+                )
+                await update.message.reply_document(
+                    document=io.BytesIO(full_output.encode()),
+                    filename=filename,
+                    caption=f"📊 Comparación: {topic_a} vs {topic_b}"
+                )
+                try:
+                    await msg.delete()
+                except Exception:
+                    pass
+            else:
+                await msg.edit_text(full_output, parse_mode=ParseMode.MARKDOWN)
+
+        except asyncio.CancelledError:
+            await msg.edit_text("🛑 Comparación cancelada.")
+        except Exception as e:
+            logger.error("Compare task failed", error=str(e))
+            try:
+                await msg.edit_text(f"❌ Error: {str(e)[:200]}")
+            except Exception:
+                pass
+        finally:
+            await db_conn_a.close()
+            await db_conn_b.close()
+
+    task = asyncio.create_task(run_compare())
+    _active_tasks[chat_id] = task
+
+
 def create_bot() -> Application:
-    app = Application.builder().token(settings.telegram_bot_token).build()
+    app = (
+        Application.builder()
+        .token(settings.telegram_bot_token)
+        .post_init(_on_startup)
+        .build()
+    )
 
     app.add_handler(CommandHandler("start", cmd_start))
     app.add_handler(CommandHandler("help", cmd_help))
@@ -524,8 +1183,16 @@ def create_bot() -> Application:
     app.add_handler(CommandHandler("generate", cmd_generate))
     app.add_handler(CommandHandler("sources", cmd_sources))
     app.add_handler(CommandHandler("outputs", cmd_outputs))
+    app.add_handler(CommandHandler("export", cmd_export))
+    app.add_handler(CommandHandler("costs", cmd_costs))
+    app.add_handler(CommandHandler("watch", cmd_watch))
+    app.add_handler(CommandHandler("unwatch", cmd_unwatch))
+    app.add_handler(CommandHandler("watches", cmd_watches))
     app.add_handler(CommandHandler("process", cmd_process))
     app.add_handler(CommandHandler("cancel", cmd_cancel))
+    app.add_handler(CommandHandler("purge", cmd_purge))
+    app.add_handler(CommandHandler("publish", cmd_publish))
+    app.add_handler(CommandHandler("compare", cmd_compare))
 
     return app
 

src/config.py

@@ -31,7 +31,16 @@ class Settings(BaseSettings):
     # Processing
     chunk_size: int = Field(800, env="CHUNK_SIZE")                  # tokens per chunk
     chunk_overlap: int = Field(100, env="CHUNK_OVERLAP")
-    quality_threshold: float = Field(0.5, env="QUALITY_THRESHOLD")  # 0-1, chunks below discarded
+    quality_threshold: float = Field(0.3, env="QUALITY_THRESHOLD")  # 0-1, chunks below discarded
+
+    # Ghost CMS
+    ghost_url: Optional[str] = Field(None, env="GHOST_URL")
+    ghost_api_key: Optional[str] = Field(None, env="GHOST_API_KEY")
+    ghost_url_en: str = Field("", env="GHOST_URL_EN")
+    ghost_api_key_en: str = Field("", env="GHOST_API_KEY_EN")
+
+    # Alerts
+    cost_alert_threshold: float = Field(0.15, env="COST_ALERT_THRESHOLD")
 
     # App
     log_level: str = Field("INFO", env="LOG_LEVEL")

src/db/database.py

@@ -5,8 +5,12 @@ from pathlib import Path
 from typing import Optional
 from enum import Enum
 
+import structlog
+
 from src.config import settings
 
+logger = structlog.get_logger()
+
 
 class ResearchStatus(str, Enum):
     RUNNING = "running"
@@ -20,6 +24,9 @@ class OutputType(str, Enum):
     BLOG = "blog"
     REPORT = "report"
     THREAD = "thread"
+    REPORT_EXTENDED = "report_extended"
+    BLOG_EXTENDED = "blog_extended"
+    PODCAST_EXTENDED = "podcast_extended"
 
 
 SCHEMA = """
@@ -84,6 +91,29 @@ CREATE INDEX IF NOT EXISTS idx_sources_session ON sources(session_id);
 CREATE INDEX IF NOT EXISTS idx_chunks_session ON chunks(session_id);
 CREATE INDEX IF NOT EXISTS idx_chunks_quality ON chunks(session_id, quality_score DESC);
 CREATE INDEX IF NOT EXISTS idx_source_contents ON source_contents(source_id);
+
+CREATE TABLE IF NOT EXISTS api_usage (
+    id INTEGER PRIMARY KEY AUTOINCREMENT,
+    session_id INTEGER REFERENCES research_sessions(id),
+    call_type TEXT NOT NULL,
+    model TEXT NOT NULL,
+    input_tokens INTEGER NOT NULL,
+    output_tokens INTEGER NOT NULL,
+    cost_usd REAL NOT NULL,
+    created_at REAL NOT NULL
+);
+
+CREATE TABLE IF NOT EXISTS watched_topics (
+    id INTEGER PRIMARY KEY AUTOINCREMENT,
+    topic TEXT NOT NULL,
+    chat_id INTEGER NOT NULL,
+    interval_hours INTEGER NOT NULL DEFAULT 24,
+    next_run_at REAL NOT NULL,
+    last_run_at REAL,
+    enabled INTEGER NOT NULL DEFAULT 1,
+    created_at REAL NOT NULL,
+    UNIQUE(topic, chat_id)
+);
 """
 
 
@@ -121,6 +151,35 @@ class ResearchDB:
         row = await cursor.fetchone()
         return dict(row) if row else None
 
+    async def get_latest_session(self, chat_id: int) -> Optional[dict]:
+        cursor = await self.db.execute(
+            "SELECT * FROM research_sessions WHERE telegram_chat_id = ? ORDER BY created_at DESC LIMIT 1",
+            (chat_id,)
+        )
+        row = await cursor.fetchone()
+        return dict(row) if row else None
+
+    async def get_session_urls(self, session_id: int) -> set:
+        async with self.db.execute(
+            "SELECT url FROM sources WHERE session_id = ?", (session_id,)
+        ) as cur:
+            rows = await cur.fetchall()
+            return {r[0] for r in rows}
+
+    async def get_previous_session(self, chat_id: int, topic: str,
+                                    exclude_session_id: int) -> Optional[dict]:
+        async with self.db.execute(
+            """SELECT id, topic, status, created_at FROM research_sessions
+               WHERE telegram_chat_id = ? AND topic = ? AND id != ?
+               ORDER BY created_at DESC LIMIT 1""",
+            (chat_id, topic, exclude_session_id)
+        ) as cur:
+            row = await cur.fetchone()
+            if not row:
+                return None
+            return {"id": row[0], "topic": row[1],
+                    "status": row[2], "created_at": row[3]}
+
     async def get_active_session(self, chat_id: int) -> Optional[dict]:
         cursor = await self.db.execute(
             """SELECT * FROM research_sessions 
@@ -259,6 +318,19 @@ class ResearchDB:
         row = await cursor.fetchone()
         return row[0] if row else None
 
+    async def get_cached_content(self, url: str,
+                                  max_age_days: int = 7) -> Optional[str]:
+        threshold = time.time() - (max_age_days * 86400)
+        async with self.db.execute(
+            """SELECT sc.content FROM source_contents sc
+               JOIN sources s ON s.id = sc.source_id
+               WHERE s.url = ? AND sc.created_at > ?
+               ORDER BY sc.created_at DESC LIMIT 1""",
+            (url, threshold)
+        ) as cur:
+            row = await cur.fetchone()
+            return row[0] if row else None
+
     async def get_outputs(self, session_id: int) -> list[dict]:
         cursor = await self.db.execute(
             "SELECT * FROM outputs WHERE session_id = ? ORDER BY created_at DESC",
@@ -266,3 +338,124 @@ class ResearchDB:
         )
         rows = await cursor.fetchall()
         return [dict(r) for r in rows]
+
+    # --- API Usage ---
+
+    async def log_api_call(self, session_id, call_type: str, model: str,
+                            input_tokens: int, output_tokens: int):
+        # Precios Claude Haiku (claude-haiku-4-5):
+        # input: $0.80 / 1M tokens  output: $4.00 / 1M tokens
+        cost = (input_tokens * 0.80 + output_tokens * 4.00) / 1_000_000
+        await self.db.execute(
+            """INSERT INTO api_usage
+               (session_id, call_type, model, input_tokens, output_tokens, cost_usd, created_at)
+               VALUES (?,?,?,?,?,?,?)""",
+            (session_id, call_type, model, input_tokens, output_tokens, cost, time.time())
+        )
+        await self.db.commit()
+
+    async def get_usage_stats(self, session_id: int) -> list[dict]:
+        cursor = await self.db.execute(
+            """SELECT call_type,
+                      COUNT(*) as calls,
+                      SUM(input_tokens + output_tokens) as total_tokens,
+                      SUM(cost_usd) as total_cost
+               FROM api_usage WHERE session_id = ?
+               GROUP BY call_type""",
+            (session_id,)
+        )
+        rows = await cursor.fetchall()
+        return [dict(r) for r in rows]
+
+    async def get_total_usage_stats(self) -> dict:
+        cursor = await self.db.execute(
+            """SELECT COUNT(DISTINCT session_id) as sessions,
+                      SUM(cost_usd) as total_cost
+               FROM api_usage"""
+        )
+        row = await cursor.fetchone()
+        return dict(row) if row else {"sessions": 0, "total_cost": 0}
+
+    # --- Watched Topics ---
+
+    async def add_watch(self, topic: str, chat_id: int, interval_hours: int) -> int:
+        now = time.time()
+        cursor = await self.db.execute(
+            """INSERT OR REPLACE INTO watched_topics
+               (topic, chat_id, interval_hours, next_run_at, created_at)
+               VALUES (?, ?, ?, ?, ?)""",
+            (topic, chat_id, interval_hours, now + interval_hours * 3600, now)
+        )
+        await self.db.commit()
+        return cursor.lastrowid
+
+    async def remove_watch(self, topic: str, chat_id: int) -> bool:
+        cursor = await self.db.execute(
+            "DELETE FROM watched_topics WHERE topic = ? AND chat_id = ?",
+            (topic, chat_id)
+        )
+        await self.db.commit()
+        return cursor.rowcount > 0
+
+    async def list_watches(self, chat_id: int) -> list[dict]:
+        cursor = await self.db.execute(
+            "SELECT * FROM watched_topics WHERE chat_id = ? ORDER BY created_at ASC",
+            (chat_id,)
+        )
+        rows = await cursor.fetchall()
+        return [dict(r) for r in rows]
+
+    async def get_due_watches(self) -> list[dict]:
+        cursor = await self.db.execute(
+            "SELECT * FROM watched_topics WHERE enabled = 1 AND next_run_at <= ?",
+            (time.time(),)
+        )
+        rows = await cursor.fetchall()
+        return [dict(r) for r in rows]
+
+    async def update_watch_run(self, watch_id: int):
+        cursor = await self.db.execute(
+            "SELECT interval_hours FROM watched_topics WHERE id = ?", (watch_id,)
+        )
+        row = await cursor.fetchone()
+        if not row:
+            return
+        now = time.time()
+        await self.db.execute(
+            "UPDATE watched_topics SET last_run_at = ?, next_run_at = ? WHERE id = ?",
+            (now, now + row[0] * 3600, watch_id)
+        )
+        await self.db.commit()
+
+    # --- Maintenance ---
+
+    async def purge_old_sessions(self, max_age_days: int = 30) -> dict:
+        await self.db.execute("PRAGMA foreign_keys = ON")
+
+        threshold = time.time() - max_age_days * 86400
+        cursor = await self.db.execute(
+            "SELECT id FROM research_sessions WHERE created_at < ? AND status != 'running'",
+            (threshold,)
+        )
+        session_ids = [row[0] for row in await cursor.fetchall()]
+
+        counts = {"sessions": 0, "sources": 0, "chunks": 0, "outputs": 0}
+
+        for sid in session_ids:
+            await self.db.execute(
+                "DELETE FROM source_contents WHERE source_id IN (SELECT id FROM sources WHERE session_id = ?)",
+                (sid,)
+            )
+            cur = await self.db.execute("DELETE FROM chunks WHERE session_id = ?", (sid,))
+            counts["chunks"] += cur.rowcount
+            cur = await self.db.execute("DELETE FROM outputs WHERE session_id = ?", (sid,))
+            counts["outputs"] += cur.rowcount
+            cur = await self.db.execute("DELETE FROM sources WHERE session_id = ?", (sid,))
+            counts["sources"] += cur.rowcount
+            cur = await self.db.execute("DELETE FROM research_sessions WHERE id = ?", (sid,))
+            counts["sessions"] += cur.rowcount
+
+        await self.db.commit()
+        logger.info("Purged sessions older than days",
+                    sessions=counts["sessions"], days=max_age_days)
+        return counts

src/generator/generator.py

@@ -1,76 +1,57 @@
 """
 ResearchOwl Generators
-Produces structured outputs from processed research using Ollama
+Produces structured outputs from processed research using Claude or Ollama
 """
+import base64
+import hashlib
+import hmac
+import json
+import re
+import time
+
 import structlog
 
+from src.config import settings
 from src.processor.processor import OllamaClient, ContentProcessor
 from src.db.database import ResearchDB, OutputType
 
 logger = structlog.get_logger()
 
 PODCAST_SYSTEM = (
-    "You are a podcast scriptwriter. Write exactly as a host SPEAKS — contractions, "
+    "Escribe SIEMPRE en español. "
-    "incomplete sentences, natural pauses, rhetorical questions. "
+    "Eres un guionista de podcast. Escribe exactamente como un presentador HABLA — contracciones, "
-    "NEVER repeat a sentence, phrase, or idea you already wrote. "
+    "frases naturales, pausas, preguntas retóricas. "
-    "Each paragraph must introduce NEW information. "
+    "NUNCA repitas una frase o idea que ya escribiste. "
-    "Use [PAUSE], [EMPHASIS], [MUSIC CUE] markers sparingly."
+    "Cada párrafo debe introducir información NUEVA. "
+    "Usa marcadores [PAUSA], [ÉNFASIS], [MÚSICA] con moderación."
 )
 
 BLOG_SYSTEM = (
-    "You are a journalist writing a blog post. Use clear markdown headings. "
+    "Escribe SIEMPRE en español. "
-    "NEVER repeat the same fact or phrase twice — if you said something, move on. "
+    "Eres un periodista escribiendo un artículo de blog. Usa encabezados markdown claros. "
-    "Each section must add new information not covered in previous sections."
+    "NUNCA repitas el mismo dato o frase dos veces — si ya lo dijiste, avanza. "
+    "Cada sección debe añadir información nueva no cubierta en secciones anteriores."
 )
 
-REPORT_SYSTEM = (
+BLOG_SYSTEM_EN = (
-    "You are a research analyst. Write a structured factual report. "
+    "You write ALWAYS in English. "
-    "Be concise — do NOT pad with redundant summaries. "
+    "You are a journalist writing a blog article. Use clear markdown headers. "
-    "NEVER restate a finding already listed. Each numbered finding must be distinct."
+    "NEVER repeat the same fact or phrase twice — if you said it, move on. "
+    "Each section must add new information not covered in other sections."
 )
 
-THREAD_SYSTEM = (
+BLOG_PROMPT_EN = """\
-    "You write Twitter/X threads. Each tweet must be under 280 chars. "
+Write a blog article about: "{topic}"
-    "NEVER repeat information from a previous tweet. "
-    "Each tweet must reveal something NEW. Number them 1/N, 2/N..."
-)
-
-
-PROMPTS = {
-    OutputType.PODCAST: """\
-Write a podcast script about: "{topic}"
-
-RULES — follow strictly:
-- Write as SPOKEN WORD: contractions, natural rhythm, as if talking to a friend
-- DO NOT use formal headings like "SEGMENT 1:" — just flow naturally
-- Each paragraph must introduce a NEW fact or angle — never restate something already said
-- If you find yourself repeating, stop and jump to the next new point
-- Aim for 800-1200 words of actual spoken content
-
-STRUCTURE (use natural transitions, not headers):
-1. Hook: open with the most surprising or dramatic fact
-2. Background: how did we get here?
-3. The key evidence or events (pick the 3 most interesting)
-4. Controversy or debate around the topic
-5. What does this mean / what happened next
-
-RESEARCH MATERIAL:
-{context}
-
-Write the script now (spoken word only, no stage directions except occasional [PAUSE]):""",
-
-    OutputType.BLOG: """\
-Write a blog post about: "{topic}"
 
 RULES — follow strictly:
 - Each section under a heading must add NEW information not covered elsewhere
 - Do NOT summarize previous sections at the start of each new section
 - Do NOT repeat facts — if a fact appears once, do not mention it again
 - Use concrete details, numbers, names — avoid vague generalities
-- Target 1000-1500 words
+- Target: 1000-1500 words
 
 STRUCTURE:
-# [Compelling headline]
+# [Impactful headline]
 
 [Hook paragraph — the most surprising fact]
 
@@ -78,58 +59,301 @@ STRUCTURE:
 [Context — what, when, who — only facts not covered elsewhere]
 
 ## Key Facts
-[The most significant findings — each bullet must be distinct]
+[Most significant findings — each point must be distinct]
 
 ## Analysis / Significance
-[What this means — no repetition of Key Facts section]
+[What this means — without repeating the Key Facts section]
 
 ## Conclusion
-[Takeaway — no more than 2 sentences summarizing, then a forward-looking statement]
+[No more than 2 sentences summarizing, then a forward-looking statement]
 
 RESEARCH MATERIAL:
 {context}
 
-Write the complete blog post in markdown:""",
+Write the complete article in markdown:"""
+
+REPORT_SYSTEM = (
+    "Escribe SIEMPRE en español. "
+    "Eres un analista de investigación. Escribe un informe estructurado y factual. "
+    "Sé conciso — NO rellenes con resúmenes redundantes. "
+    "NUNCA repitas un hallazgo ya listado. Cada hallazgo numerado debe ser distinto."
+)
+
+THREAD_SYSTEM = (
+    "Escribe SIEMPRE en español. "
+    "Escribes hilos de Twitter/X. Cada tweet debe tener menos de 280 caracteres. "
+    "NUNCA repitas información de un tweet anterior. "
+    "Cada tweet debe revelar algo NUEVO. Numéralos 1/N, 2/N..."
+)
+
+
+PROMPTS = {
+    OutputType.PODCAST: """\
+Escribe un guion de podcast sobre: "{topic}"
+
+REGLAS — sigue estrictamente:
+- Escribe como PALABRA HABLADA: contracciones, ritmo natural, como si hablaras con un amigo
+- NO uses encabezados formales como "SEGMENTO 1:" — fluye de forma natural
+- Cada párrafo debe introducir un NUEVO hecho o ángulo — nunca repitas algo ya dicho
+- Si te encuentras repitiendo, para y salta al siguiente punto nuevo
+- Objetivo: 800-1200 palabras de contenido hablado real
+
+ESTRUCTURA (usa transiciones naturales, no encabezados):
+1. Gancho: abre con el hecho más sorprendente o dramático
+2. Contexto: ¿cómo llegamos aquí?
+3. Las evidencias o eventos clave (elige los 3 más interesantes)
+4. La controversia o debate sobre el tema
+5. ¿Qué significa esto / qué pasó después?
+
+MATERIAL DE INVESTIGACIÓN:
+{context}
+
+Escribe el guion ahora (solo palabra hablada, sin acotaciones excepto [PAUSA] ocasional):""",
+
+    OutputType.BLOG: """\
+Escribe un artículo de blog sobre: "{topic}"
+
+REGLAS — sigue estrictamente:
+- Cada sección bajo un encabezado debe añadir información NUEVA no cubierta en otro lugar
+- NO resumas secciones anteriores al inicio de cada nueva sección
+- NO repitas hechos — si un hecho aparece una vez, no lo menciones de nuevo
+- Usa detalles concretos, números, nombres — evita generalidades vagas
+- Objetivo: 1000-1500 palabras
+
+ESTRUCTURA:
+# [Titular impactante]
+
+[Párrafo gancho — el hecho más sorprendente]
+
+## Contexto
+[Contexto — qué, cuándo, quién — solo hechos no cubiertos en otro lugar]
+
+## Hechos Clave
+[Los hallazgos más significativos — cada punto debe ser distinto]
+
+## Análisis / Importancia
+[Qué significa esto — sin repetir la sección de Hechos Clave]
+
+## Conclusión
+[Conclusión — no más de 2 oraciones resumiendo, luego una declaración prospectiva]
+
+MATERIAL DE INVESTIGACIÓN:
+{context}
+
+Escribe el artículo completo en markdown:""",
 
     OutputType.REPORT: """\
-Write a research report about: "{topic}"
+Escribe un informe de investigación sobre: "{topic}"
 
-RULES — follow strictly:
+REGLAS — sigue estrictamente:
-- Each numbered finding must be DISTINCT — no overlapping content
+- Cada hallazgo numerado debe ser DISTINTO — sin contenido que se superponga
-- The Executive Summary must NOT repeat findings verbatim — only the 2-3 most critical points
+- El Resumen Ejecutivo NO debe repetir los hallazgos literalmente — solo los 2-3 puntos más críticos
-- Source quality and contradictions must reference specific claims, not generic statements
+- La calidad de las fuentes y contradicciones deben referenciar afirmaciones específicas, no declaraciones genéricas
-- Be precise and concise — no filler
+- Sé preciso y conciso — sin relleno
 
-STRUCTURE:
+ESTRUCTURA:
-1. Executive Summary (3-4 sentences, key takeaways only)
+1. Resumen Ejecutivo (3-4 oraciones, solo puntos clave)
-2. Key Findings (5-10 numbered, each completely distinct)
+2. Hallazgos Clave (5-10 numerados, cada uno completamente distinto)
-3. Evidence Analysis (what the sources show, with any contradictions)
+3. Análisis de Evidencia (lo que muestran las fuentes, con cualquier contradicción)
-4. Timeline (if applicable — specific dates/events)
+4. Cronología (si aplica — fechas/eventos específicos)
-5. Conclusions & Open Questions
+5. Conclusiones y Preguntas Abiertas
 
-RESEARCH MATERIAL:
+MATERIAL DE INVESTIGACIÓN:
 {context}
 
-Write the complete report in markdown:""",
+Escribe el informe completo en markdown:""",
 
     OutputType.THREAD: """\
-Write a Twitter/X thread about: "{topic}"
+Escribe un hilo de Twitter/X sobre: "{topic}"
 
-RULES — follow strictly:
+REGLAS — sigue estrictamente:
-- Each tweet must reveal ONE new fact or idea — never restate a previous tweet
+- Cada tweet debe revelar UN nuevo hecho o idea — nunca repetir un tweet anterior
-- Max 280 characters per tweet (count carefully)
+- Máximo 280 caracteres por tweet (cuenta cuidadosamente)
-- Number format: 1/ 2/ 3/ ... N/
+- Formato de numeración: 1/ 2/ 3/ ... N/
-- Hook tweet must be the most surprising/provocative fact
+- El tweet gancho debe ser el hecho más sorprendente/provocador
-- Build toward a conclusion — do not repeat the hook at the end
+- Avanza hacia una conclusión — no repitas el gancho al final
-- 12-18 tweets total
+- 12-18 tweets en total
 
-RESEARCH MATERIAL:
+MATERIAL DE INVESTIGACIÓN:
 {context}
 
-Write the thread (one tweet per line, nothing else):"""
+Escribe el hilo (un tweet por línea, nada más):"""
 }
 
 
+OUTLINE_REPORT = """
+IMPORTANTE: Los títulos y queries deben estar en español, independientemente del idioma del material de investigación.
+
+Eres un editor de investigación. Dado el tema "{topic}" y el material
+disponible, genera un outline detallado para un informe exhaustivo.
+
+Devuelve SOLO una lista JSON de secciones, sin texto adicional, sin
+markdown, sin explicaciones. Formato exacto:
+[
+  {{"title": "Título de la sección", "query": "términos de búsqueda específicos para esta sección", "words": 800}},
+  ...
+]
+
+Genera entre 6 y 10 secciones. Cada sección debe:
+- Cubrir un ángulo distinto del tema
+- Tener una query específica para recuperar chunks relevantes
+- Indicar longitud objetivo en palabras (400-1200)
+
+Material disponible (resumen):
+{context_summary}
+"""
+
+OUTLINE_BLOG = """
+IMPORTANTE: Los títulos y queries deben estar en español, independientemente del idioma del material de investigación.
+
+Eres un editor de contenido. Dado el tema "{topic}" y el material
+disponible, genera un outline para un artículo de blog exhaustivo.
+
+Devuelve SOLO una lista JSON de secciones, sin texto adicional:
+[
+  {{"title": "Título de sección", "query": "términos búsqueda", "words": 600}},
+  ...
+]
+
+Genera entre 5 y 8 secciones. Primera sección = introducción gancho.
+Última sección = conclusión con perspectiva original.
+
+Material disponible (resumen):
+{context_summary}
+"""
+
+OUTLINE_PODCAST = """
+IMPORTANTE: Los títulos y queries deben estar en español, independientemente del idioma del material de investigación.
+
+Eres un productor de podcast. Dado el tema "{topic}" y el material
+disponible, genera un outline para un guion de podcast exhaustivo.
+
+Devuelve SOLO una lista JSON de segmentos, sin texto adicional:
+[
+  {{"title": "Nombre del segmento", "query": "términos búsqueda", "words": 700}},
+  ...
+]
+
+Genera entre 5 y 7 segmentos. Flujo natural: gancho → contexto →
+desarrollo → controversia → conclusión.
+
+Material disponible (resumen):
+{context_summary}
+"""
+
+
+# ─── Ghost CMS ────────────────────────────────────────────────────────────────
+
+def _b64url(data: bytes | str) -> str:
+    if isinstance(data, str):
+        data = data.encode()
+    return base64.urlsafe_b64encode(data).rstrip(b"=").decode()
+
+
+def _extract_title(content: str) -> str:
+    """Return first H1 heading from markdown, skipping the ResearchOwl header block."""
+    in_header = False
+    for line in content.splitlines():
+        stripped = line.strip()
+        if stripped == "---":
+            in_header = not in_header
+            continue
+        if in_header:
+            continue
+        if stripped.startswith("# ") and not stripped.startswith("## "):
+            return stripped[2:].strip()
+    return ""
+
+
+def _strip_researchowl_header(content: str) -> str:
+    """Remove the ---...--- metadata block that ResearchOwl prepends to outputs."""
+    lines = content.splitlines(keepends=True)
+    dashes_seen = 0
+    for i, line in enumerate(lines):
+        if line.strip() == "---":
+            dashes_seen += 1
+            if dashes_seen == 2:
+                return "".join(lines[i + 1:]).lstrip("\n")
+    return content
+
+
+class GhostPublisher:
+    def __init__(self, lang: str = "es"):
+        if lang == "en":
+            self.url = (settings.ghost_url_en or "").rstrip("/")
+            self.api_key = settings.ghost_api_key_en or ""
+        else:
+            self.url = (settings.ghost_url or "").rstrip("/")
+            self.api_key = settings.ghost_api_key or ""
+
+    def is_configured(self) -> bool:
+        return bool(self.url and self.api_key)
+
+    def _make_token(self) -> str:
+        key_id, secret = self.api_key.split(":", 1)
+        now = int(time.time())
+        header = _b64url(json.dumps({"alg": "HS256", "typ": "JWT", "kid": key_id}))
+        payload = _b64url(json.dumps({"iat": now, "exp": now + 300, "aud": "/admin/"}))
+        signing = f"{header}.{payload}"
+        sig = _b64url(
+            hmac.new(bytes.fromhex(secret), signing.encode(), hashlib.sha256).digest()
+        )
+        return f"{signing}.{sig}"
+
+    async def publish_draft(self, title: str, markdown_content: str,
+                            tags: list[str] | None = None) -> dict:
+        import aiohttp as _aio
+        import markdown as _md
+
+        clean = _strip_researchowl_header(markdown_content)
+        html = _md.markdown(clean, extensions=["extra"])
+
+        # Ghost añade el título automáticamente — eliminar el primer <h1> para evitar duplicado
+        html = re.sub(r"<h1[^>]*>.*?</h1>", "", html, count=1, flags=re.DOTALL).lstrip()
+
+        logger.info("Ghost publish_draft", html_length=len(html),
+                    html_preview=html[:200])
+
+        if not html.strip():
+            raise ValueError("Ghost: HTML vacío tras conversión markdown — contenido no enviado")
+
+        # Ghost 5.x (Lexical editor): el campo "html" es solo de lectura en la API.
+        # La forma fiable de enviar HTML arbitrario es via mobiledoc con HTML card,
+        # que Ghost acepta en todas las versiones de v5 y renderiza sin conversión.
+        mobiledoc = json.dumps({
+            "version": "0.3.1",
+            "atoms": [],
+            "cards": [["html", {"html": html}]],
+            "markups": [],
+            "sections": [[10, 0]],
+        })
+
+        token = self._make_token()
+        body = {
+            "posts": [{
+                "title": title,
+                "mobiledoc": mobiledoc,
+                "status": "draft",
+                "tags": [{"name": t} for t in (tags or ["investigacion"])],
+            }]
+        }
+        async with _aio.ClientSession() as sess:
+            async with sess.post(
+                f"{self.url}/ghost/api/admin/posts/",
+                json=body,
+                headers={
+                    "Authorization": f"Ghost {token}",
+                    "Accept-Version": "v5.0",
+                },
+            ) as resp:
+                if resp.status not in (200, 201):
+                    text = await resp.text()
+                    raise ValueError(f"Ghost API {resp.status}: {text[:300]}")
+                return await resp.json()
+
+
+# ─── Output generation ────────────────────────────────────────────────────────
+
 class OutputGenerator:
     def __init__(self, db: ResearchDB, ollama: OllamaClient, processor: ContentProcessor):
         self.db = db
@@ -137,8 +361,14 @@ class OutputGenerator:
         self.processor = processor
 
     async def generate(self, session_id: int, output_type: OutputType,
-                       progress_callback=None) -> str:
+                       progress_callback=None, lang: str = "es") -> str:
         """Generate an output for a research session"""
+        if output_type in (OutputType.REPORT_EXTENDED,
+                           OutputType.BLOG_EXTENDED,
+                           OutputType.PODCAST_EXTENDED):
+            return await self.generate_extended(session_id, output_type, progress_callback,
+                                                lang=lang)
+
         session = await self.db.get_session(session_id)
         if not session:
             raise ValueError(f"Session {session_id} not found")
@@ -151,7 +381,7 @@ class OutputGenerator:
 
         # RAG: get most relevant context for this output type
         query = self._get_rag_query(output_type, topic)
-        context = await self.processor.rag_query(session_id, query, top_k=30)
+        context = await self.processor.rag_query(session_id, query, top_k=80)
 
         if not context:
             # Fallback: use raw top chunks
@@ -161,20 +391,19 @@ class OutputGenerator:
         if not context:
             raise ValueError("No processed content available. Run /process first.")
 
-        # Truncate context to avoid Ollama context limits
+        backend = "Claude Haiku" if settings.anthropic_api_key else "Ollama"
-        context_words = context.split()
-        if len(context_words) > 6000:
-            context = " ".join(context_words[:6000]) + "\n\n[... additional material truncated ...]"
-
         if progress_callback:
-            await progress_callback(f"✍️ Generating {output_type} with Ollama... (this takes 2-5 min)")
+            await progress_callback(f"✍️ Generando {output_type} con {backend}... (2-5 min)")
 
         # Build prompt
         system = self._get_system(output_type)
         prompt = PROMPTS[output_type].format(topic=topic, context=context)
 
-        # Generate — temperature=0.7 reduces repetition in small models
+        if lang == "en" and output_type == OutputType.BLOG:
-        output = await self.ollama.generate(prompt, system=system, timeout=300, temperature=0.7)
+            system = BLOG_SYSTEM_EN
+            prompt = BLOG_PROMPT_EN.format(topic=topic, context=context)
+
+        output = await self._generate(prompt, system, output_type, session_id)
 
         # Add metadata header
         stats = await self.db.get_session_stats(session_id)
@@ -184,8 +413,60 @@ class OutputGenerator:
         # Save to DB
         await self.db.save_output(session_id, output_type, full_output)
 
+        # Auto-publish to Ghost for blog outputs
+        ghost_notice = ""
+        if output_type in (OutputType.BLOG, OutputType.BLOG_EXTENDED):
+            ghost = GhostPublisher(lang=lang)
+            if ghost.is_configured():
+                try:
+                    title = _extract_title(full_output) or topic
+                    result = await ghost.publish_draft(title, full_output)
+                    post = result["posts"][0]
+                    ghost_notice = (
+                        f"\n\n---\n"
+                        f"📤 *Borrador publicado en Ghost*\n"
+                        f"Editar: {ghost.url}/ghost/#/editor/post/{post['id']}"
+                    )
+                    logger.info("Auto-published blog to Ghost", post_id=post["id"])
+                except Exception as e:
+                    logger.warning("Auto-publish to Ghost failed", error=str(e))
+
         logger.info("Output generated", type=output_type, length=len(full_output))
-        return full_output
+        return full_output + ghost_notice
+
+    async def _generate(self, prompt: str, system: str, output_type: OutputType,
+                         session_id: int | None = None) -> str:
+        if settings.anthropic_api_key:
+            return await self._generate_with_claude(prompt, system, output_type, session_id)
+        return await self._generate_with_ollama(prompt, system)
+
+    async def _generate_with_claude(self, prompt: str, system: str, output_type: OutputType,
+                                     session_id: int | None = None) -> str:
+        import anthropic
+        max_tokens = 4096 if output_type == OutputType.THREAD else 16000
+        try:
+            client = anthropic.AsyncAnthropic(api_key=settings.anthropic_api_key)
+            msg = await client.messages.create(
+                model=settings.claude_model,
+                max_tokens=max_tokens,
+                system=system,
+                messages=[{"role": "user", "content": prompt}],
+            )
+            if session_id is not None:
+                try:
+                    await self.db.log_api_call(
+                        session_id, "generation", settings.claude_model,
+                        msg.usage.input_tokens, msg.usage.output_tokens
+                    )
+                except Exception as log_err:
+                    logger.warning("Failed to log API usage", error=str(log_err))
+            return msg.content[0].text.strip()
+        except Exception as e:
+            logger.warning("Claude generation failed, falling back to Ollama", error=str(e))
+            return await self._generate_with_ollama(prompt, system)
+
+    async def _generate_with_ollama(self, prompt: str, system: str) -> str:
+        return await self.ollama.generate(prompt, system=system, timeout=300, temperature=0.7)
 
     def _get_rag_query(self, output_type: OutputType, topic: str) -> str:
         queries = {
@@ -205,6 +486,155 @@ class OutputGenerator:
         }
         return systems.get(output_type, "You are a helpful research assistant.")
 
+    async def generate_extended(self, session_id: int, output_type: OutputType,
+                                progress_callback=None, lang: str = "es") -> str:
+        """
+        Generación por secciones para outputs exhaustivos.
+        1. Recupera muestra de contexto para el outline
+        2. Genera outline con Claude (lista de secciones)
+        3. Para cada sección: RAG específico → genera sección
+        4. Concatena y guarda
+        """
+        session = await self.db.get_session(session_id)
+        if not session:
+            raise ValueError(f"Session {session_id} not found")
+        topic = session["topic"]
+
+        # Paso 1: contexto resumen para el outline (top 10 chunks)
+        top_chunks = await self.db.get_top_chunks(session_id, limit=10)
+        if not top_chunks:
+            raise ValueError("No processed content available. Run /process first.")
+        context_summary = "\n\n".join(
+            f"- {c.get('title', '')}: {c['content'][:300]}"
+            for c in top_chunks
+        )
+
+        if progress_callback:
+            await progress_callback("🗂️ Generando estructura del documento…")
+
+        # Paso 2: outline
+        base_type = output_type.value.replace("_extended", "")
+        outline_prompts = {
+            "report": OUTLINE_REPORT,
+            "blog": OUTLINE_BLOG,
+            "podcast": OUTLINE_PODCAST,
+        }
+        outline_prompt = outline_prompts[base_type].format(
+            topic=topic, context_summary=context_summary
+        )
+
+        outline_json = await self._generate_raw(outline_prompt, session_id)
+        try:
+            import json as _json
+            clean = outline_json.strip()
+            if clean.startswith("```"):
+                clean = "\n".join(clean.split("\n")[1:])
+            if clean.endswith("```"):
+                clean = "\n".join(clean.split("\n")[:-1])
+            sections = _json.loads(clean.strip())
+        except Exception as e:
+            logger.error("Failed to parse outline", error=str(e), raw=outline_json[:200])
+            raise ValueError(f"No se pudo generar el outline: {e}")
+
+        if progress_callback:
+            await progress_callback(
+                f"✍️ Generando {len(sections)} secciones… (esto tardará varios minutos)"
+            )
+
+        # Paso 3: generar cada sección
+        base_output_type = OutputType(base_type)
+        system = self._get_system(base_output_type)
+        if lang == "en" and output_type == OutputType.BLOG_EXTENDED:
+            system = BLOG_SYSTEM_EN
+        sections_text = []
+
+        for i, section in enumerate(sections, 1):
+            title = section.get("title", f"Sección {i}")
+            query = section.get("query", topic)
+            target_words = section.get("words", 600)
+
+            if progress_callback:
+                await progress_callback(
+                    f"✍️ Sección {i}/{len(sections)}: {title[:40]}…"
+                )
+
+            section_context = await self.processor.rag_query(session_id, query, top_k=40)
+            if not section_context:
+                section_context = context_summary
+
+            lang_rule = "- Write in English\n" if lang == "en" else "- Escribe en español\n"
+            section_prompt = (
+                f"Escribe la sección '{title}' del {base_type} sobre: '{topic}'\n\n"
+                f"REGLAS:\n"
+                f"- NO incluyas ningún título o encabezado al inicio de tu respuesta — el título de la sección ya está incluido externamente\n"
+                f"- Esta es UNA sección de un documento más largo — no repitas introducción ni conclusión general\n"
+                f"- No incluyas encabezados del documento completo, solo el contenido de esta sección\n"
+                f"- Objetivo: aproximadamente {target_words} palabras\n"
+                f"- Usa SOLO información del material siguiente — no inventes datos\n"
+                f"{lang_rule}"
+                f"\nMATERIAL:\n{section_context}"
+            )
+
+            section_text = await self._generate(
+                section_prompt, system, base_output_type, session_id
+            )
+            sections_text.append(f"## {title}\n\n{section_text}")
+
+        # Paso 4: concatenar
+        full_content = "\n\n---\n\n".join(sections_text)
+        stats = await self.db.get_session_stats(session_id)
+        header = self._build_header(topic, output_type, session, stats)
+        full_output = header + "\n\n" + full_content
+
+        await self.db.save_output(session_id, output_type, full_output)
+
+        # Auto-publish to Ghost for extended blog outputs
+        ghost_notice = ""
+        if output_type == OutputType.BLOG_EXTENDED:
+            ghost = GhostPublisher(lang=lang)
+            if ghost.is_configured():
+                try:
+                    title = _extract_title(full_output) or topic
+                    result = await ghost.publish_draft(title, full_output)
+                    post = result["posts"][0]
+                    ghost_notice = (
+                        f"\n\n---\n"
+                        f"📤 *Borrador publicado en Ghost*\n"
+                        f"Editar: {ghost.url}/ghost/#/editor/post/{post['id']}"
+                    )
+                    logger.info("Auto-published extended blog to Ghost", post_id=post["id"])
+                except Exception as e:
+                    logger.warning("Auto-publish to Ghost failed (extended)", error=str(e))
+
+        logger.info("Extended output generated", type=output_type,
+                    sections=len(sections), length=len(full_output))
+        return full_output + ghost_notice
+
+    async def _generate_raw(self, prompt: str,
+                             session_id: int | None = None) -> str:
+        if settings.anthropic_api_key:
+            import anthropic
+            try:
+                client = anthropic.AsyncAnthropic(api_key=settings.anthropic_api_key)
+                msg = await client.messages.create(
+                    model=settings.claude_model,
+                    max_tokens=2048,
+                    messages=[{"role": "user", "content": prompt}],
+                )
+                if session_id is not None:
+                    try:
+                        await self.db.log_api_call(
+                            session_id, "outline", settings.claude_model,
+                            msg.usage.input_tokens, msg.usage.output_tokens
+                        )
+                    except Exception:
+                        pass
+                return msg.content[0].text.strip()
+            except Exception as e:
+                logger.warning("Claude outline failed", error=str(e))
+                raise
+        raise ValueError("Claude API key required for extended generation")
+
     def _build_header(self, topic: str, output_type: OutputType,
                       session: dict, stats: dict) -> str:
         from datetime import datetime
@@ -218,3 +648,279 @@ Iterations: {session.get('iterations', 0)}
 Total words researched: {session.get('total_words', 0):,}
 ---
 """
+
+
+def _remove_duplicate_headings(text: str) -> str:
+    lines = text.split('\n')
+    result = []
+    i = 0
+    while i < len(lines):
+        line = lines[i].rstrip()
+        if line.startswith('# ') and not line.startswith('## '):
+            h1_text = line[2:].strip().lower()
+            window = result[-5:] if len(result) >= 5 else result[:]
+            recent_h2s = {
+                p.strip()[3:].strip().lower()
+                for p in window
+                if p.strip().startswith('## ')
+            }
+            if h1_text in recent_h2s:
+                i += 1
+                continue
+        result.append(lines[i])
+        i += 1
+    return '\n'.join(result)
+
+
+def generate_pdf(content: str, title: str = "ResearchOwl Output") -> bytes:
+    content = _remove_duplicate_headings(content)
+    try:
+        from reportlab.lib.pagesizes import A4
+        from reportlab.lib.styles import getSampleStyleSheet, ParagraphStyle
+        from reportlab.lib.units import cm
+        from reportlab.platypus import SimpleDocTemplate, Paragraph, Spacer, HRFlowable
+        from reportlab.lib.enums import TA_LEFT
+        from reportlab.lib import colors
+        import io
+        import re
+    except ImportError:
+        raise ImportError("reportlab is required for PDF export — pip install reportlab")
+
+    buf = io.BytesIO()
+    doc = SimpleDocTemplate(
+        buf,
+        pagesize=A4,
+        rightMargin=2 * cm,
+        leftMargin=2 * cm,
+        topMargin=2.5 * cm,
+        bottomMargin=2 * cm,
+        title=title,
+    )
+
+    base = getSampleStyleSheet()
+    normal = ParagraphStyle("RO_Normal", parent=base["Normal"],
+                            fontSize=10, leading=14, spaceAfter=4)
+    h1 = ParagraphStyle("RO_H1", parent=base["Heading1"],
+                        fontSize=18, spaceBefore=12, spaceAfter=6,
+                        textColor=colors.HexColor("#1a1a2e"))
+    h2 = ParagraphStyle("RO_H2", parent=base["Heading2"],
+                        fontSize=14, spaceBefore=10, spaceAfter=4,
+                        textColor=colors.HexColor("#16213e"))
+    h3 = ParagraphStyle("RO_H3", parent=base["Heading3"],
+                        fontSize=12, spaceBefore=8, spaceAfter=4)
+    code_style = ParagraphStyle("RO_Code", parent=base["Code"],
+                                fontSize=9, leading=12, fontName="Courier",
+                                backColor=colors.HexColor("#f4f4f4"), spaceAfter=4)
+    bullet_style = ParagraphStyle("RO_Bullet", parent=normal,
+                                  leftIndent=20, bulletIndent=10, spaceAfter=2)
+
+    def md_to_para(text: str) -> str:
+        text = text.replace("&", "&amp;").replace("<", "&lt;").replace(">", "&gt;")
+        text = re.sub(r'\*\*(.+?)\*\*', r'<b>\1</b>', text)
+        text = re.sub(r'__(.+?)__', r'<b>\1</b>', text)
+        text = re.sub(r'\*(.+?)\*', r'<i>\1</i>', text)
+        text = re.sub(r'_(.+?)_', r'<i>\1</i>', text)
+        text = re.sub(r'`(.+?)`', r'<font name="Courier">\1</font>', text)
+        return text
+
+    story = []
+    lines = content.split("\n")
+    in_code = False
+    code_buf = []
+
+    for line in lines:
+        if line.startswith("```"):
+            if not in_code:
+                in_code = True
+                code_buf = []
+            else:
+                in_code = False
+                try:
+                    story.append(Paragraph(
+                        "<br/>".join(l.replace("&", "&amp;").replace("<", "&lt;").replace(">", "&gt;")
+                                     for l in code_buf),
+                        code_style
+                    ))
+                except Exception:
+                    pass
+            continue
+
+        if in_code:
+            code_buf.append(line)
+            continue
+
+        if re.match(r'^[-*_]{3,}$', line.strip()):
+            story.append(HRFlowable(width="100%", thickness=0.5,
+                                    color=colors.grey, spaceAfter=6))
+            continue
+
+        try:
+            if line.startswith("### "):
+                story.append(Paragraph(md_to_para(line[4:]), h3))
+            elif line.startswith("## "):
+                story.append(Paragraph(md_to_para(line[3:]), h2))
+            elif line.startswith("# "):
+                story.append(Paragraph(md_to_para(line[2:]), h1))
+            elif re.match(r'^[-*+] ', line):
+                story.append(Paragraph("• " + md_to_para(line[2:]), bullet_style))
+            elif re.match(r'^\d+\. ', line):
+                story.append(Paragraph(md_to_para(line), bullet_style))
+            elif line.strip() == "":
+                story.append(Spacer(1, 6))
+            else:
+                story.append(Paragraph(md_to_para(line), normal))
+        except Exception:
+            try:
+                story.append(Paragraph(line[:300], normal))
+            except Exception:
+                pass
+
+    doc.build(story)
+    return buf.getvalue()
+
+
+async def generate_diff_summary(
+    topic: str,
+    new_urls: set,
+    old_urls: set,
+    new_chunks: list,
+    session_id: int,
+    db,
+) -> str | None:
+    from src.config import settings
+    import structlog
+    diff_logger = structlog.get_logger()
+
+    added_urls = new_urls - old_urls
+    pct_new = len(added_urls) / max(len(new_urls), 1)
+
+    diff_logger.info("Diff analysis", topic=topic,
+                     new_urls=len(new_urls), old_urls=len(old_urls),
+                     added=len(added_urls), pct_new=round(pct_new, 2))
+
+    if pct_new < 0.20 and len(added_urls) < 5:
+        diff_logger.info("Diff: no significant new sources", topic=topic)
+        return None
+
+    if not new_chunks:
+        return None
+
+    context = "\n\n---\n\n".join(
+        f"[{c.get('source_type', 'web').upper()}] {c.get('title', '')}\n{c['content'][:400]}"
+        for c in new_chunks[:20]
+    )
+
+    if not settings.anthropic_api_key:
+        return (
+            f"📊 *Novedades detectadas sobre {topic}*\n\n"
+            f"• {len(added_urls)} fuentes nuevas encontradas\n"
+            f"• {len(new_chunks)} chunks de contenido procesados\n\n"
+            f"Usa /generate report para ver el análisis completo."
+        )
+
+    try:
+        import anthropic
+        client = anthropic.AsyncAnthropic(api_key=settings.anthropic_api_key)
+        prompt = (
+            f'Analiza el siguiente material de investigación sobre "{topic}" '
+            f'y genera un resumen BREVE (máximo 300 palabras) de las novedades '
+            f'más importantes encontradas. Escribe en español.\n\n'
+            f'Si el contenido es muy similar a investigaciones anteriores o no '
+            f'contiene información genuinamente nueva, responde SOLO con: '
+            f'"SIN_NOVEDADES"\n\n'
+            f'Material nuevo:\n{context}'
+        )
+        msg = await client.messages.create(
+            model=settings.claude_model,
+            max_tokens=500,
+            messages=[{"role": "user", "content": prompt}]
+        )
+        summary = msg.content[0].text.strip()
+
+        if summary == "SIN_NOVEDADES":
+            diff_logger.info("Diff: Claude found no new information", topic=topic)
+            return None
+
+        try:
+            await db.log_api_call(session_id, "diff", settings.claude_model,
+                                  msg.usage.input_tokens, msg.usage.output_tokens)
+        except Exception:
+            pass
+
+        return f"🔔 *Novedades — {topic}*\n\n{summary}\n\nUsa /generate para report completo."
+
+    except Exception as e:
+        diff_logger.warning("Diff summary generation failed", error=str(e))
+        return (
+            f"📊 *Actualización — {topic}*\n\n"
+            f"• {len(added_urls)} fuentes nuevas\n"
+            f"Usa /generate report para ver el análisis completo."
+        )
+
+
+async def generate_comparison(
+    topic_a: str,
+    topic_b: str,
+    context_a: str,
+    context_b: str,
+    session_id_a: int,
+    db,
+) -> str:
+    from src.config import settings
+    import structlog
+    cmp_logger = structlog.get_logger()
+
+    if not settings.anthropic_api_key:
+        raise ValueError("Claude API key required for comparison")
+
+    def _truncate(text: str, max_words: int = 3000) -> str:
+        words = text.split()
+        if len(words) > max_words:
+            return " ".join(words[:max_words]) + "\n\n[... contenido adicional truncado ...]"
+        return text
+
+    context_a = _truncate(context_a)
+    context_b = _truncate(context_b)
+
+    prompt = (
+        f'Eres un analista experto. Compara en profundidad estos dos temas:\n'
+        f'TEMA A: "{topic_a}"\n'
+        f'TEMA B: "{topic_b}"\n\n'
+        f'Escribe el análisis en español con esta estructura:\n\n'
+        f'## Resumen comparativo\n'
+        f'(2-3 párrafos con las diferencias y similitudes más importantes)\n\n'
+        f'## {topic_a}\n'
+        f'(Puntos clave únicos de este tema)\n\n'
+        f'## {topic_b}\n'
+        f'(Puntos clave únicos de este tema)\n\n'
+        f'## Similitudes\n'
+        f'(Qué tienen en común)\n\n'
+        f'## Diferencias clave\n'
+        f'(Tabla markdown o lista de las diferencias más relevantes)\n\n'
+        f'## Conclusión\n'
+        f'(Cuál es mejor/más relevante según el contexto, o qué conclusión se extrae)\n\n'
+        f'---\n'
+        f'MATERIAL DE INVESTIGACIÓN — {topic_a}:\n{context_a}\n\n'
+        f'---\n'
+        f'MATERIAL DE INVESTIGACIÓN — {topic_b}:\n{context_b}\n'
+    )
+
+    try:
+        import anthropic
+        client = anthropic.AsyncAnthropic(api_key=settings.anthropic_api_key)
+        msg = await client.messages.create(
+            model=settings.claude_model,
+            max_tokens=8192,
+            messages=[{"role": "user", "content": prompt}]
+        )
+        try:
+            await db.log_api_call(
+                session_id_a, "comparison", settings.claude_model,
+                msg.usage.input_tokens, msg.usage.output_tokens
+            )
+        except Exception:
+            pass
+        return msg.content[0].text.strip()
+    except Exception as e:
+        cmp_logger.error("Comparison generation failed", error=str(e))
+        raise

src/processor/processor.py

@@ -129,6 +129,8 @@ class ContentProcessor:
         scraped = [s for s in sources if s["status"] == "scraped"]
 
         logger.info("Processing sources", total=len(scraped))
+        scraped = await self._dedup_sources(session_id, scraped)
+        logger.info("After dedup", unique=len(scraped))
         total_chunks = 0
         total_words = 0
 
@@ -161,6 +163,56 @@ class ContentProcessor:
 
         return {"total_chunks": total_chunks, "total_words": total_words}
 
+    async def _dedup_sources(self, session_id: int,
+                              scraped: list[dict]) -> list[dict]:
+        try:
+            import hashlib
+            seen_hashes: set = set()
+            seen_prefixes: list = []
+            unique: list = []
+            duplicates = 0
+
+            for source in scraped:
+                content = await self.db.get_source_content(source["id"])
+                if not content:
+                    unique.append(source)
+                    continue
+
+                content_hash = hashlib.md5(content[:2000].encode()).hexdigest()
+                if content_hash in seen_hashes:
+                    duplicates += 1
+                    await self.db.update_source(source["id"], status="skipped")
+                    continue
+                seen_hashes.add(content_hash)
+
+                prefix = content[:300].strip().lower()
+                prefix_words = set(prefix.split())
+                is_dup = False
+                if len(prefix_words) >= 10:
+                    for seen_prefix_words in seen_prefixes:
+                        intersection = len(prefix_words & seen_prefix_words)
+                        union = len(prefix_words | seen_prefix_words)
+                        if intersection / max(union, 1) > 0.85:
+                            is_dup = True
+                            break
+
+                if is_dup:
+                    duplicates += 1
+                    await self.db.update_source(source["id"], status="skipped")
+                    continue
+
+                seen_prefixes.append(prefix_words)
+                unique.append(source)
+
+            if duplicates > 0:
+                logger.info("Dedup complete", session_id=session_id,
+                            original=len(scraped), duplicates=duplicates,
+                            unique=len(unique))
+            return unique
+        except Exception as e:
+            logger.warning("Dedup failed, processing all sources", error=str(e))
+            return scraped
+
     async def _process_source(self, session_id: int, topic: str, source: dict) -> int:
         """Chunk, score, embed and store a single source. Returns chunk count."""
         source_id = source["id"]
@@ -182,7 +234,7 @@ class ContentProcessor:
             if words < 30:
                 continue
 
-            quality = await self._score_quality(chunk, topic)
+            quality = await self._score_quality(chunk, topic, session_id)
             if quality < settings.quality_threshold:
                 filtered_quality += 1
                 logger.debug("Chunk filtered by quality", source_id=source_id,
@@ -215,16 +267,20 @@ class ContentProcessor:
         logger.info("Source processed", source_id=source_id, stored=stored)
         return stored
 
-    async def _score_quality(self, chunk: str, topic: str) -> float:
+    async def _score_quality(self, chunk: str, topic: str,
+                              session_id: int | None = None) -> float:
         """Score 0-1 relevance to topic. Uses Claude Haiku if API key set, else Ollama."""
         if settings.anthropic_api_key:
-            return await self._score_with_claude(chunk, topic)
+            return await self._score_with_claude(chunk, topic, session_id)
         return await self._score_with_ollama(chunk, topic)
 
-    async def _score_with_claude(self, chunk: str, topic: str) -> float:
+    async def _score_with_claude(self, chunk: str, topic: str,
+                                  session_id: int | None = None) -> float:
         import anthropic
         prompt = (
             f'Rate 0-10 how relevant this text is to the topic "{topic}". '
+            f'Be generous — if the text is tangentially related, score 4+. '
+            f'Only score below 3 if completely unrelated. '
             f'Reply with only a number.\n\nText:\n{chunk[:500]}'
         )
         try:
@@ -234,6 +290,14 @@ class ContentProcessor:
                 max_tokens=10,
                 messages=[{"role": "user", "content": prompt}]
             )
+            if session_id is not None:
+                try:
+                    await self.db.log_api_call(
+                        session_id, "scoring", settings.claude_model,
+                        msg.usage.input_tokens, msg.usage.output_tokens
+                    )
+                except Exception as log_err:
+                    logger.warning("Failed to log API usage", error=str(log_err))
             response = msg.content[0].text.strip()
             numbers = re.findall(r'\b(\d+(?:\.\d+)?)\b', response)
             if numbers:
@@ -241,7 +305,7 @@ class ContentProcessor:
                 normalized = min(1.0, score / 10.0)
                 logger.debug("Claude relevance score", raw=score, normalized=round(normalized, 2))
                 return normalized
-            return 0.6
+            return 0.5
         except Exception as e:
             logger.warning("Claude scoring failed, falling back to Ollama", error=str(e))
             return await self._score_with_ollama(chunk, topic)
@@ -275,7 +339,7 @@ class ContentProcessor:
         query_embedding = await self.ollama.embed(query)
 
         # Get top quality chunks
-        chunks = await self.db.get_top_chunks(session_id, limit=100)
+        chunks = await self.db.get_top_chunks(session_id, limit=300)
 
         if query_embedding and chunks:
             # Rank by embedding similarity

src/scraper/exhaustive.py

@@ -3,6 +3,7 @@ ResearchOwl Exhaustive Scraper
 Core engine: discovers, expands, and evaluates sources recursively
 """
 import asyncio
+import random
 import re
 import time
 from typing import Optional
@@ -54,6 +55,22 @@ REDDIT_RE = re.compile(r"reddit\.com/(r/\w+/comments/\w+)")
 WIKIPEDIA_RE = re.compile(r"wikipedia\.org/wiki/(.+)")
 
 
+async def fetch_with_retry(fetch_fn, source_name: str, max_retries: int = 3):
+    last_exc = None
+    for attempt in range(max_retries):
+        try:
+            return await fetch_fn()
+        except Exception as e:
+            last_exc = e
+            if attempt < max_retries - 1:
+                wait = 2 ** attempt + random.random()
+                logger.debug("fetch_with_retry backoff", source=source_name[:60],
+                             attempt=attempt + 1, wait=round(wait, 1), error=str(e))
+                await asyncio.sleep(wait)
+    logger.warning("fetch_with_retry exhausted", source=source_name[:60], error=str(last_exc))
+    raise last_exc
+
+
 def detect_source_type(url: str) -> str:
     if YOUTUBE_RE.search(url):
         return "youtube"
@@ -140,16 +157,15 @@ class ExhaustiveScraper:
         """Initial broad search across multiple sources"""
         logger.info("Seeding research", topic=self.topic)
         tasks = [
-            self._seed_duckduckgo(),
+            self._seed_search(),
             self._seed_wikipedia(),
             self._seed_reddit(),
             self._seed_youtube(),
         ]
         await asyncio.gather(*tasks, return_exceptions=True)
 
-    async def _seed_duckduckgo(self):
+    async def _generate_ddg_queries(self) -> list[str]:
-        """Multiple DDG queries — fresh DDGS() per query to avoid cascading ratelimits"""
+        fallback = [
-        queries = [
             self.topic,
             f"{self.topic} history facts",
             f"{self.topic} evidence analysis",
@@ -159,13 +175,103 @@ class ExhaustiveScraper:
             f"{self.topic} documentary",
             f"{self.topic} research study",
         ]
+
+        if not settings.anthropic_api_key:
+            return fallback
+
+        try:
+            import anthropic
+            logger.info("Generating DDG queries with Claude", topic=self.topic)
+            client = anthropic.AsyncAnthropic(api_key=settings.anthropic_api_key)
+            prompt = (
+                f'Generate exactly 8 DuckDuckGo search queries to research: "{self.topic}"\n\n'
+                f'Rules:\n'
+                f'- Each query must be specific and distinct — no generic templates\n'
+                f'- Cover different angles: facts, history, official sources, criticism, '
+                f'technical details, recent developments, expert opinions, primary sources\n'
+                f'- Use the most specific terminology for this topic\n'
+                f'- Include the topic language naturally (if topic is in Spanish, '
+                f'mix Spanish and English queries for broader coverage)\n'
+                f'- Output ONLY the 8 queries, one per line, no numbering, '
+                f'no explanations, no markdown\n'
+            )
+            msg = await client.messages.create(
+                model=settings.claude_model,
+                max_tokens=300,
+                messages=[{"role": "user", "content": prompt}]
+            )
+            raw = msg.content[0].text.strip()
+            queries = [q.strip() for q in raw.split('\n') if q.strip()]
+            if self.topic not in queries:
+                queries = [self.topic] + queries[:7]
+            queries = queries[:8]
+            logger.info("DDG queries generated by Claude", queries=queries)
+            return queries
+        except Exception as e:
+            logger.warning("Claude query generation failed, using fallback",
+                           error=str(e), error_type=type(e).__name__)
+            return fallback
+
+    async def _search_searxng(self, query: str) -> list[dict]:
+        """Busca en SearXNG y retorna lista de {href, title}. Retorna [] si no disponible."""
+        import aiohttp
+        searxng_url = "http://searxng-svc.researchowl.svc.cluster.local:8080/search"
+        params = {
+            "q": query,
+            "format": "json",
+            "engines": "duckduckgo,google,bing,brave",
+            "language": "all",
+        }
+        headers = {
+            "Accept": "application/json",
+            "X-Forwarded-For": "127.0.0.1",
+            "User-Agent": "ResearchOwl/1.0",
+        }
+        try:
+            async with aiohttp.ClientSession() as session:
+                async with session.get(
+                    searxng_url,
+                    params=params,
+                    headers=headers,
+                    timeout=aiohttp.ClientTimeout(total=15)
+                ) as resp:
+                    if resp.status == 200:
+                        data = await resp.json()
+                        results = data.get("results", [])
+                        logger.info("SearXNG query ok", query=query, results=len(results))
+                        return [
+                            {"href": r.get("url", ""), "title": r.get("title", "")}
+                            for r in results
+                            if r.get("url")
+                        ]
+                    else:
+                        logger.warning("SearXNG non-200", status=resp.status, query=query)
+                        return []
+        except Exception as e:
+            logger.warning("SearXNG failed", query=query, error=str(e))
+            return []
+
+    async def _seed_search(self):
+        """SearXNG primary + DDG fallback per query"""
+        queries = await self._generate_ddg_queries()
         for query in queries:
             if self._stop:
                 break
+            results = await self._search_searxng(query)
+            if not results:
+                logger.info("SearXNG vacío, usando DDG", query=query)
                 try:
-                # Fresh instance per query — a ratelimit on one won't poison the rest
                     with DDGS() as ddgs:
-                    results = list(ddgs.text(query, max_results=settings.max_pages_per_search))
+                        ddg_results = list(ddgs.text(
+                            query,
+                            max_results=settings.max_pages_per_search
+                        ))
+                    results = ddg_results
+                    logger.info("DDG fallback ok", query=query, results=len(results))
+                except Exception as e:
+                    logger.warning("DDG fallback failed", query=query, error=str(e))
+                    results = []
+
             for r in results:
                 url = normalize_url(r.get("href", ""))
                 if url and not is_blacklisted(url):
@@ -175,10 +281,7 @@ class ExhaustiveScraper:
                         depth=0,
                         title=r.get("title")
                     )
-                logger.info("DDG query ok", query=query, results=len(results))
+            await asyncio.sleep(random.uniform(1, 3))
-            except Exception as e:
-                logger.warning("DDG query failed", query=query, error=str(e))
-            await asyncio.sleep(settings.request_delay * 2)
 
     async def _seed_wikipedia(self):
         """Search Wikipedia API for correct article URLs.
@@ -290,12 +393,7 @@ class ExhaustiveScraper:
             )
 
             if self.progress_callback:
-                await self.progress_callback(
+                await self.progress_callback(self.iteration, self.total_sources)
-                    iteration=self.iteration,
-                    total=self.total_sources,
-                    new_this_round=new_sources,
-                    stats=stats
-                )
 
             # Saturation check: if we found very few new URLs, we're done
             if new_sources < 3 and self.iteration > 2:
@@ -316,10 +414,31 @@ class ExhaustiveScraper:
             source_id = source["id"]
 
             try:
+                try:
+                    cached = await self.db.get_cached_content(url)
+                except Exception as cache_err:
+                    logger.warning("Cache lookup failed", url=url, error=str(cache_err))
+                    cached = None
+
+                if cached:
+                    logger.debug("Cache hit", url=url)
+                    await self.db.save_source_content(source_id, cached)
+                    await self.db.update_source(
+                        source_id,
+                        status="scraped",
+                        scraped_at=time.time(),
+                        word_count=len(cached.split()),
+                    )
+                    return 0
+
                 if source_type == "youtube":
-                    content, title = await self._extract_youtube(url)
+                    content, title = await fetch_with_retry(
+                        lambda: self._extract_youtube(url), url
+                    )
                 elif source_type == "wikipedia":
-                    content, title, new_urls = await self._extract_wikipedia(url)
+                    content, title, new_urls = await fetch_with_retry(
+                        lambda: self._extract_wikipedia(url), url
+                    )
                     added = 0
                     for new_url in (new_urls or []):
                         if self._url_is_relevant(new_url):
@@ -331,13 +450,18 @@ class ExhaustiveScraper:
                     await self._mark_scraped(source_id, content, title, url)
                     return added
                 elif source_type == "reddit":
-                    content, title = await self._extract_reddit(url)
+                    content, title = await fetch_with_retry(
-                    # Small delay between Reddit requests to avoid rate limiting
+                        lambda: self._extract_reddit(url), url
+                    )
                     await asyncio.sleep(settings.request_delay)
                 elif source_type == "pdf":
-                    content, title = await self._extract_pdf(url)
+                    content, title = await fetch_with_retry(
+                        lambda: self._extract_pdf(url), url
+                    )
                 else:
-                    content, title, new_urls = await self._extract_web(url, source["depth"])
+                    content, title, new_urls = await fetch_with_retry(
+                        lambda: self._extract_web(url, source["depth"]), url
+                    )
                     added = 0
                     for new_url in (new_urls or []):
                         if self._url_is_relevant(new_url):
@@ -469,12 +593,23 @@ class ExhaustiveScraper:
             return None, None
 
         video_id = match.group(1)
-        try:
+        loop = asyncio.get_event_loop()
-            transcript_list = YouTubeTranscriptApi.get_transcript(
+
+        def _fetch():
+            return YouTubeTranscriptApi.get_transcript(
                 video_id, languages=["en", "es", "en-US", "en-GB"]
             )
+
+        try:
+            transcript_list = await asyncio.wait_for(
+                loop.run_in_executor(None, _fetch),
+                timeout=30.0
+            )
             text = " ".join(t["text"] for t in transcript_list)
             return text, f"YouTube: {video_id}"
+        except asyncio.TimeoutError:
+            logger.warning("YouTube transcript timed out", video_id=video_id)
+            return None, None
         except NoTranscriptFound:
             return None, None
         except Exception as e: