Content Gen Skill — v1.2 (Quality-Gated)

Generate a full month of content in 2 hours, or get a single piece in 30 seconds — with a neural quality score on every output.

What Changed in v1.2

Previous versions generated content and stopped. v1.2 adds a content-scorer quality gate:

1. 1. Content is generated
2. Output is scored against 8 neuromarketing dimensions (0-100)
3. If score is below your threshold (default: 70), the skill identifies the weakest dimensions and regenerates with specific improvement instructions
4. Final output includes the neural score as a deliverable proof point

This means you never receive content that hasn't passed a quality check — and your customers see the score.

What This Skill Does

Bulk or single-piece content for:

• - 30-day social calendars — full month of posts across X, LinkedIn, Facebook, Instagram
• Video scripts — short-form (30s/60s) and long-form (3-5 min)
• SEO articles — keyword-targeted blog posts
• Listing copy — property descriptions from address or specs
• Agent partnership decks — talking points, one-pagers, pitch scripts
• Objection handling scripts — responses for common buyer/seller objections
• Market update newsletters — weekly/monthly digest for lead nurture lists

Input Contract

CODEBLOCK0

Output Contract

All outputs now include a neural_score block:

CODEBLOCK1

The Quality Gate — How It Works

Neural Scoring Dimensions (8 factors, total 100pts)

Auto-Rewrite Logic

CODEBLOCK2

Default threshold: 70/100 (above industry average of 60)
Recommended threshold for paid products: 75+
Max rewrites: 2 (prevents infinite loops)

Modes

Calendar:
CODEBLOCK3

Script:
CODEBLOCK4

Article: Full markdown with meta description, keyword density note, and per-section score.

All other modes: Markdown with score badge appended.

Score Badge (Use as Social Proof)

Every output includes a ready-to-use score badge. Add it to your Gumroad product page, Fiverr delivery, or social post:

"This content scored 79/100 on neural signal strength — above the 60/100 industry average for mortgage marketing content."

Buyers don't need to understand the methodology. They see proof it's been tested.

Example — Quality Gate in Action

Input: "Write a LinkedIn post about VA loans for veterans who are renting"

Draft 1 score: 61 (below 70 threshold)

• - Weak: second-person density (8/20) — uses "veterans" not "you"
• Weak: specificity (1/4) — no dollar amounts or timelines
• Auto-improvement: switch to direct second-person, add "$0 down" and "no PMI" specifics

Draft 2 score: 78 (above threshold, delivered)

• - "If you served and you're renting right now — you may be leaving a real advantage on the table. VA loans let eligible veterans buy with zero down and no PMI — saving $150–$300/month compared to a conventional loan on the same home."

Output includes: final content + 78/100 score + "1 rewrite needed" note

Backend

Uses generate.py (shared backend). Local MLX first, Haiku fallback.

CODEBLOCK5

Note: Scoring runs locally (no additional API cost). Only generation uses LLM.

Tiers

Free — Single-piece generation only, no auto-rewrite, score reported but no gate
Standard ($12/mo) — All modes, quality gate active, up to 1 rewrite, score badge included
Pro ($37/mo) — All modes, up to 2 rewrites, threshold configurable, bulk calendar with per-post scores, structured JSON export

Integration

CODEBLOCK6