blog-pipeline

Blog Pipeline

Reference document showing the full Blog Engine flow with human approval gates. This is not an executable command — it's a map of how blog commands connect.

Trigger

User invokes /blog-pipeline or says "show me the blog flow", "how does the blog engine work", "blog pipeline diagram"

Pipeline Overview

BLOG ENGINE — Full Pipeline
════════════════════════════════════════════════════════

  PLAN
  ────

  /blog-plan [topic]
    |
    |  1. Load content strategy / article plan
    |  2. Research canon sources ({{CANON_DIR}})
    |  3. Build structured outline with section-level sources
    |  4. Draft visual map (cover + inline + CTA)
    |  5. Draft CMS metadata
    |  6. Canon traceability check (every claim sourced)
    |
    |  Output: outline.md with sections, sources, visual map
    |
    |  [HUMAN GATE 1: Review and approve outline]
    |  ├── "go" ──> proceed to draft
    |  └── feedback ──> revise outline, re-present
    |
    v

  DRAFT
  ─────

  /blog-draft [topic]
    |
    |  1. Load approved outline
    |  2. Load brand/canon-rules.json
    |  3. Load content-engine/memory/lessons.md
    |  4. Write full article section by section
    |  5. Run /stop-slop (remove AI writing patterns)
    |  6. Insert visual placeholders at marked positions
    |  7. Verify word count targets per section
    |
    |  Output: draft.md — full article with visual markers
    |
    |  [HUMAN GATE 2: Review full draft]
    |  ├── "approved" ──> proceed to assets
    |  └── feedback ──> revise sections, re-present
    |
    v

  ASSETS
  ──────

  /blog-assets [topic]
    |
    |  1. Read visual map from outline
    |  2. Read draft for context
    |  3. Create each asset:
    |     ├── Cover image (1200x675, dark theme default)
    |     ├── Inline visuals (tables, flows, comparisons)
    |     └── CTA card (standard template)
    |  4. Export as PNG to {{CONTENT_OUTPUT_DIR}}/blog/[slug]/assets/
    |
    |  Output: PNG files + asset manifest
    |
    |  [HUMAN GATE 3: Review visual assets]
    |  ├── "approved" ──> proceed to QA
    |  └── feedback ──> revise specific assets
    |
    v

  QA
  ──

  /blog-qa [topic]
    |
    |  Checks:
    |  ├── Canon compliance
    |  │   ├── Banned phrases (from canon-rules.json)
    |  │   ├── Required language (terminology enforcement)
    |  │   └── USP alignment (only approved USPs)
    |  │
    |  ├── /compliance-check (regulatory scan)
    |  │   ├── Jurisdiction-specific rules
    |  │   ├── Missing disclaimers
    |  │   └── Financial promotion flags
    |  │
    |  ├── Source verification
    |  │   ├── Every claim traced to canon source
    |  │   ├── Single-source claims flagged
    |  │   └── Unverified claims marked
    |  │
    |  ├── Structure check
    |  │   ├── Word count within target range
    |  │   ├── H2 count (target 5-7)
    |  │   ├── Visual placement (cover + CTA minimum)
    |  │   └── CMS metadata complete
    |  │
    |  └── Brand voice
    |      ├── Tone consistency
    |      ├── No AI writing patterns
    |      └── Audience alignment
    |
    |  Result: PASS or FAIL with specific violations
    |
    ├── PASS ──> proceed to publish
    |
    └── FAIL ──> Fix violations
                   |
                   v
                 Re-run /blog-qa (loop until PASS)
                   |
                   v
                 /lesson-capture (auto-capture failure pattern)

    |
    v

  PUBLISH
  ───────

  /blog-publish [topic]
    |
    |  [HUMAN GATE 4: Final approval before CMS push]
    |
    |  1. Verify QA passed (refuse if last QA was FAIL)
    |  2. Prepare CMS payload:
    |     ├── Copy (plain markdown or HTML)
    |     ├── Cover image
    |     ├── Inline images
    |     ├── CTA card
    |     └── CMS metadata (title, excerpt, category, date)
    |  3. Push to CMS (Framer, WordPress, Ghost, etc.)
    |  4. Verify publish (check live URL)
    |
    |  Output: live URL + confirmation
    |
    v
  Done. Article is live.

════════════════════════════════════════════════════════