blog: calibration v8 — AI phrasing blacklist, STEP8 6-step rewrite, Flexoptix author identity

- STRICTLY FORBIDDEN section: comprehensive AI phrasing blacklist (leverage, utilize,
  Furthermore, robust, seamless, delve into, in conclusion, etc.)
- STEP8 rewritten into 6 structured steps: hunt AI words, break rhythm, add human element,
  fix label formats, structural cleanup, ensure Flexoptix identity
- Banned sentence structures added (parallel triplets, same-length paragraphs)
- STEP6 author identity: reader should know this came from Flexoptix
- Version bump to 0.3.0

packages/api/src/index.ts

@@ -69,7 +69,7 @@ app.get("/", (_req, res) => {
 app.get("/api", (_req, res) => {
   res.json({
     name: "Transceiver Intelligence Platform",
-    version: "0.2.6",
+    version: "0.3.0",
     endpoints: [
       "GET /api/transceivers?q=&form_factor=&speed=&category=&fiber_type=&wdm_type=&coherent=",
       "GET /api/transceivers/:id",

packages/api/src/llm/fo-blog-pipeline.ts

@@ -42,9 +42,39 @@ VOICE:
 - Occasionally blunt or sarcastic
 - Prioritize clarity over completeness
 
-STRICTLY FORBIDDEN:
+STRICTLY FORBIDDEN — AI PHRASING BLACKLIST (any of these = rewrite):
+
+BANNED WORDS & PHRASES (AI fingerprints — never use):
+- "leverage", "optimize", "enhance", "plays a key role", "unlock", "empower"
+- "In today's fast-paced world", "In the ever-evolving landscape", "it goes without saying"
+- "it is worth noting", "it is important to note", "it should be noted"
+- "that being said", "with that in mind", "at the end of the day"
+- "delve into", "dive into", "explore", "unpack", "shed light on"
+- "game-changer", "game-changing", "cutting-edge", "state-of-the-art", "industry-leading"
+- "robust", "seamless", "scalable", "holistic", "comprehensive" (as empty adjectives)
+- "it is crucial to", "it is essential to", "it is vital to"
+- "in conclusion", "to summarize", "in summary", "to wrap up"
+- "Furthermore", "Additionally", "Moreover", "Consequently", "Subsequently"
+- "This is why X is not optional, but part of the baseline operating model" — McKinsey speak
+- "X rarely comes from a single obvious source" — vague academic hedge
+- "The discussion around X is often framed as" — consulting opening
+- "In practice, things tend to behave differently" — too soft, say HOW they behave
+- "a testament to", "a reflection of", "underscores the importance of"
+- "revolutionary", "industry-leading", "next-generation", "world-class"
+- "streamline", "streamlined", "best-in-class", "cutting edge"
+- "nuanced", "multifaceted", "ecosystem" (when used vaguely)
+- "paradigm", "synergy", "utilize" (say "use")
+
+BANNED SENTENCE STRUCTURES (AI patterns):
+- Perfectly parallel sentences: "X does A. Y does B. Z does C." — vary the rhythm
+- Every paragraph same length — break it
+- Sentences starting with "This" three times in a row — AI hallmark
+- Ending a section with a 3-4 word summary sentence that restates what was just said
+- "Both X and Y have their place" — wishy-washy non-conclusion
+- "Ultimately, the decision depends on..." without saying what it depends on specifically
+
+BANNED STRUCTURAL PATTERNS:
 - "In today's fast-paced world" or ANY generic intro
-- "leverage", "optimize", "enhance", "plays a key role"
 - Empty bullet lists without context
 - Neutral non-advice ("it depends on your requirements")
 - Textbook explanations of basic concepts
@@ -332,33 +362,66 @@ Article:
 // STEP 8: KILL AI TONE
 // ═══════════════════════════════════════════════════════
 
-export const STEP8_KILL_AI_TONE = `Rewrite this article to remove ALL signs of AI-generated text.
+export const STEP8_KILL_AI_TONE = `Your task: make this article sound like it was written by a human engineer who has actual opinions — not by a language model trying to be balanced and comprehensive.
 
-REMOVE:
+STEP 1 — HUNT AND DESTROY AI WORDS (scan every sentence, replace or delete):
-- Overly perfect sentence structures
+These words are AI fingerprints. Any of them = rewrite that sentence.
-- Repetitive paragraph patterns (same opening, same length)
+→ "leverage" → use "use"
-- Generic transition phrases ("Furthermore", "Additionally", "It's worth noting")
+→ "utilize" → use "use"
-- Lists that all follow identical format
+→ "Furthermore", "Additionally", "Moreover" → delete the word, rewrite transition naturally or cut
-- Perfect grammar everywhere — add occasional conversational shortcuts
+→ "It is worth noting", "It is important to note", "it should be noted" → delete entirely, just say the thing
-- Phrases like "it is important to note", "one should consider"
+→ "delve into", "dive into", "explore" → skip the preamble, start doing it
-- Any section that ends with a tidy 4-item bullet list
+→ "robust" as empty adjective → name what makes it robust, or delete
-- Perfectly symmetrical sections (same length = AI fingerprint)
+→ "seamless" → delete, nothing is seamless
-- ALL markdown headers: ##, ###, ####, **Bold Section Title:** — strip every single one. Write plain text only.
+→ "holistic", "comprehensive" as empty adjectives → delete
-- Repeated explanations: if cleaning, polarity, or power is in two sections, merge them into the first occurrence and remove the second.
+→ "that being said", "with that in mind", "at the end of the day" → delete
-- "Cause:" / "Fix:" / "Example:" labels — these are documentation format, not blog format. Integrate into prose.
+→ "in conclusion", "to summarize", "in summary" → delete, article already said it
-  BAD: "**Cause:** Mismatched firmware versions. **Fix:** Validate compatibility before deployment."
+→ "a testament to", "underscores the importance of", "a reflection of" → rewrite directly
-  GOOD: "Firmware mismatches are rarely obvious. In one deployment, optics initialized cleanly but started flapping under load — the issue turned out to be a version mismatch between switch and transceiver firmware that only surfaced under realistic traffic."
+→ "nuanced", "multifaceted" → say what you actually mean
-- Repeated MMF→SMF transitions: if the cabling change is explained more than once, merge into one place.
+→ "streamline", "streamlined" → say what changes specifically
-- "fraught with hidden risks" or similar fear-selling language — replace with "introduces additional operational considerations"
+→ "paradigm" → delete or say "approach"
-- "When Not to Use" as a bullet list — integrate into prose: "In smaller environments or legacy MMF deployments, the overhead of validation and migration can outweigh the cost advantage."
+→ "synergy" → delete always
-- "Example:" as a standalone label before a quote — weave the example into the surrounding paragraph naturally.
+→ "it is crucial to", "it is essential to", "it is vital to" → just say the thing without the preamble
+→ "This is why X is not optional, but part of the baseline operating model" → say "X is required. Period."
+→ "In practice, things tend to behave differently" → say HOW they behave differently, with an example
+→ "The discussion is often framed as X versus Y" → skip framing, start with the actual point
+→ "Both X and Y have their place" → take a position or cut the sentence
+→ "Ultimately" as sentence opener → delete or rephrase
 
-REPLACE WITH:
+STEP 2 — BREAK PERFECT AI RHYTHM:
-- Natural, slightly imperfect flow
+AI writes in perfectly even waves. Humans don't. Fix it:
-- Varied sentence lengths (some very short, some longer)
+→ After 3 same-length sentences: add ONE very short one. "It usually doesn't." or "That's the problem."
-- Conversational asides ("Look, ...", "Here's the thing:", "Don't get me started on...")
+→ After a long technical paragraph: one direct opinion sentence. "Most teams don't catch this until it's too late."
-- Direct address ("You know this is true if...")
+→ Vary paragraph length — some 1 sentence, some 4. Never all the same.
-- Specific instead of generic ("the Nexus 93180 in rack 14" not "your network switch")
+→ If every paragraph starts with "The" or "In" — vary the opening words
+
+STEP 3 — ADD ONE HUMAN ELEMENT PER ARTICLE:
+Choose ONE of these and add it naturally:
+→ A conversational aside: "Look, nobody is going to tell you this in a sales call, but..."
+→ A direct address: "If you've been on a pager call at 2AM for a dirty connector, you know exactly what I mean."
+→ A blunt Flexoptix statement: "We sell compatible optics. We've also seen what happens when teams skip validation. The optics aren't the problem."
+→ An admission: "There's no clean answer here. The right choice depends on whether you've done the work upfront."
+
+STEP 4 — FIX LABEL FORMATS:
+→ "Cause:" / "Fix:" / "Example:" as labels → integrate into prose narrative
+  BAD: "Cause: Firmware mismatch. Fix: Validate before deployment."
+  GOOD: "In one deployment, links came up cleanly but started flapping under load. The root cause was a firmware version gap between switch and transceiver — something that only surfaced under real traffic. Aligning firmware across both platforms fixed it."
+→ "When Not to Use" as bullet list → one flowing paragraph
+→ "BUY / WAIT / SKIP" sections → delete, say it as one direct sentence instead
+
+STEP 5 — STRUCTURAL CLEANUP:
+→ ALL markdown headers: ##, ###, #### → remove, plain text only
+→ Repeated topics (cleaning in two sections, MMF/SMF twice) → merge into first occurrence, delete second
+→ "fraught with" → replace with specific description of what happens
+→ Sentences starting "This" three times in a row → vary the opener
+→ ALL markdown headers: ##, ###, ####, **Bold Section Title:** → strip every single one
+
+STEP 6 — ENSURE FLEXOPTIX IS RECOGNIZABLE AS AUTHOR:
+→ The reader should know this came from Flexoptix — not a generic tech blog
+→ If Flexoptix appears 0 times in the article: add it once naturally
+  Example: "This is something we see regularly at Flexoptix — teams that saved 70% on optics and spent twice that on troubleshooting because they skipped the validation step."
+→ The author sells optics. They have an opinion. They've seen what fails. Show that.
+→ One sentence of Flexoptix perspective is enough — don't make it an ad, just make it human
 
 PROSE STYLE OPTION (use when article currently feels too structured/sectioned):
 If the article has many headers and bullet points and reads like a slide deck, consider

packages/api/src/routes/health.ts

@@ -14,7 +14,7 @@ healthRouter.get("/", async (_req: Request, res: Response) => {
     res.json({
       success: true,
       status: "healthy",
-      version: "0.2.6",
+      version: "0.3.0",
       uptime: process.uptime(),
       database: {
         connected: true,

packages/dashboard/hot-topics.js

@@ -21,10 +21,10 @@
     }
     blogPipelineRunning = true;
 
-    var blogList = document.getElementById('blog-list');
+    var pipelineEl = document.getElementById('blog-pipeline-status');
-    if (blogList) {
+    if (pipelineEl) {
-      blogList.innerHTML =
+      pipelineEl.innerHTML =
-        '<div style="background:linear-gradient(135deg,#1a1a1a,#2a2a2a);color:white;padding:2rem;border-radius:12px;text-align:center">' +
+        '<div style="background:linear-gradient(135deg,#1a1a1a,#2a2a2a);color:white;padding:2rem;border-radius:12px;text-align:center;margin-bottom:1rem">' +
         '<div style="font-size:1.4rem;font-weight:700;margin-bottom:1rem">Generating Blog with AI...</div>' +
         '<div id="bp-status" style="font-size:1rem;color:#FF8100;margin-bottom:0.5rem">Starting 10-step Flexoptix Style pipeline...</div>' +
         '<div id="bp-step" style="font-size:0.85rem;color:#aaa">Connecting to LLM (qwen2.5:14b)</div>' +
@@ -63,15 +63,36 @@
   function pollPipeline(id, attempt) {
     if (attempt > 80) {
       blogPipelineRunning = false;
+      var pipelineEl = document.getElementById('blog-pipeline-status');
+      if (pipelineEl) pipelineEl.innerHTML = '';
       if (typeof showToast === 'function') showToast('Timeout', 'Pipeline took too long. Check the blog list.');
       if (typeof loadBlogDrafts === 'function') loadBlogDrafts();
       return;
     }
+    // Poll progress endpoint (fast) + blog status endpoint (to detect completion)
     setTimeout(function() {
-      fetch((API || '') + '/api/blog/' + id).then(function(r) { return r.json(); }).then(function(data) {
+      // 1) Fetch real-time progress from server
+      fetch((API || '') + '/api/blog/' + id + '/progress')
+        .then(function(r) { return r.json(); })
+        .then(function(prog) {
+          if (prog.running) {
+            var bar = document.getElementById('bp-bar');
+            var pct = document.getElementById('bp-pct');
+            var status = document.getElementById('bp-status');
+            var step = document.getElementById('bp-step');
+            if (bar) bar.style.width = prog.pct + '%';
+            if (pct) pct.textContent = prog.pct + '%';
+            if (status) status.textContent = prog.label || ('Step ' + prog.step + '/10');
+            if (step) step.textContent = 'Running on qwen2.5:14b via Ollama...';
+          }
+        }).catch(function() {});
+
+      // 2) Fetch blog draft to detect pipeline completion
+      fetch((API || '') + '/api/blog/' + id)
+        .then(function(r) { return r.json(); })
+        .then(function(data) {
           var d = data.draft || data;
           var gen = d.generated_by || '';
-        var steps = d.pipeline_steps_completed || 0;
           var done = gen && gen !== 'tip-blog-engine-template' && gen.length > 0;
 
           if (done) {
@@ -82,29 +103,21 @@
             var step = document.getElementById('bp-step');
             if (bar) bar.style.width = '100%';
             if (pct) pct.textContent = '100%';
-          if (status) { status.textContent = 'Blog generated! ' + (d.word_count || '?') + ' words'; status.style.color = '#2d6a4f'; }
+            if (status) { status.textContent = '✓ Blog fertig! ' + (d.word_count || '?') + ' Wörter'; status.style.color = '#2d6a4f'; }
-          if (step) step.textContent = 'Generated by: ' + gen;
+            if (step) step.textContent = 'Engine: ' + gen;
             blogPipelineRunning = false;
             if (typeof showToast === 'function') showToast('Blog Ready!', (d.title || 'Article') + ' — ' + (d.word_count || '?') + ' words');
             setTimeout(function() {
+              var pipelineEl = document.getElementById('blog-pipeline-status');
+              if (pipelineEl) pipelineEl.innerHTML = '';
               if (typeof loadBlogDrafts === 'function') loadBlogDrafts();
               if (typeof viewBlogDraft === 'function') viewBlogDraft(id);
             }, 2000);
           } else {
-          // Still processing
-          var pctVal = Math.min(95, steps * 10 + 5);
-          var bar = document.getElementById('bp-bar');
-          var pct = document.getElementById('bp-pct');
-          var status = document.getElementById('bp-status');
-          var step = document.getElementById('bp-step');
-          if (bar) bar.style.width = pctVal + '%';
-          if (pct) pct.textContent = pctVal + '%';
-          if (status) status.textContent = 'Step ' + steps + '/10';
-          if (step && steps > 0 && steps <= 10) step.textContent = STEP_NAMES[steps - 1] || 'Processing...';
             pollPipeline(id, attempt + 1);
           }
         }).catch(function() { pollPipeline(id, attempt + 1); });
-    }, 15000);
+    }, 8000); // Poll every 8s instead of 15s for snappier UI
   }
 
   // Hot topics loader
@@ -149,6 +162,13 @@
   }
 })();
 
+// Called via data attributes to avoid quote-escaping issues in onclick
+window.blogDeleteClick = function(el) {
+  var id = el.getAttribute('data-blog-id');
+  var title = el.getAttribute('data-blog-title');
+  if (id) window.deleteBlogDraft(id, title);
+};
+
 // Delete a single blog draft
 window.deleteBlogDraft = function(id, title) {
   if (!confirm('Delete "' + title + '"?')) return;