---
title: Content Quality Gate
category: product
entity_type: skill
price: $15
canonical: https://forgehouse.ai/skills/content-quality-gate/
lang: en
hreflang_alt: https://forgehouse.ai/tr/skiller/content-quality-gate/
last_updated: 2026-06-20
---

# Content Quality Gate

> Score content quality with AI phrase detection, readability analysis, keyword density, and…

An automatic quality barrier between content production and publishing that scores any blog post, service page, or landing page across five dimensions on a 0-100 scale. It detects AI-typical phrasing, checks readability and keyword density, scrubs invisible Unicode and vague filler, and routes anything below threshold into an auto-revision loop.

## Use cases
- Quality-checking a blog post or service page before publishing
- Stripping AI-tell phrases like 'delve into' or 'seamless' from marketing copy
- Verifying keyword density stays in the natural 0.5-2.5% range
- Forcing vague claims to be replaced with concrete numbers and proof
- Cleaning invisible Unicode watermarks that leave AI-generation traces
- Running content through a checker that is separate from the writer (Maker-Checker)

## Benefits
- Stops low-quality content from reaching readers and damaging trust
- Makes content feel human and specific instead of generic AI slop
- Reduces waste by fixing only the weakest dimension instead of rewriting everything
- Protects SEO standing against helpful-content signals that penalize thin writing

## What’s included
- Five-dimension scoring with weights (humanity, specificity, structure, SEO, readability)
- A threshold system (pass / auto-revise / review-required) with a CLI analyzer
- An extensive AI-phrase, buzzword, and hedging dictionary plus replacement suggestions
- A content scrubber for invisible Unicode, em-dash rules, and 26 vague-word fixes
- Turkish-calibrated readability so the agglutinative language is scored correctly
- An auto-revision loop that targets the lowest-scoring dimension first, capped at retries

## Who it’s for
Content and SEO teams who want every published piece to pass an objective, specificity-driven quality bar before it goes live.

## How it runs
Between draft and publish sits a scored gate: five weighted dimensions, a 70-point bar, and a revise loop that fixes the weakest dimension first while scrubbing AI cliches, invisible characters and fuzzy words.
1. Runs the content-analyzer CLI on the draft with language and target keyword flags, producing a JSON composite score from 5 weighted dimensions: humanity and voice 30 percent, specificity and depth 25, structure and readability 20, SEO optimization 15, readability score 10.
2. Applies the threshold system: 70 and above passes to publish, 50 to 69 enters the auto-revise loop, below 50 stops and escalates for human review.
3. In the revise loop, fixes the single lowest-scoring dimension first instead of rewriting everything: if humanity is at 41, it removes the flagged AI phrases and replaces them with specific data, then rescores. Maximum 2 revision attempts before the content is rewritten from scratch.
4. Scrubs AI fingerprints as part of every pass: a banned-phrase dictionary of 50+ AI cliches in English and Turkish, passive-voice detection above 20 percent, keyword stuffing above 2.5 percent density, and context-aware em-dash replacement.
5. Strips invisible forensic traces: zero-width spaces, byte order marks and BIDI control characters that betray AI generation are removed with a single regex sweep.
6. Hunts vagueness with a 26-word fuzzy-word list (many, significantly, usually, nearly and friends): each hit is either deleted if the sentence survives without it, or replaced with a concrete number, measurement or dated fact.

## FAQ
### Does it score copy from any source, or only AI-generated drafts?
Any blog post, service page or landing page goes through the same five-dimension score, whoever wrote it. A human draft full of vague filler fails the same checks an AI draft would.

### A single 0-to-100 number feels arbitrary. What is actually behind it?
It is five dimensions rather than one gut call: AI-typical phrasing, readability, keyword density in the natural range, invisible Unicode and vague filler. Each is a concrete check, so the score traces back to specific issues you can fix.

### When it flags a piece, does it rewrite it or just block it?
It scrubs invisible Unicode and obvious filler, but for real issues it routes the piece back rather than silently rewriting your voice. A passing score means it cleared an objective bar, not that the argument is automatically compelling.

## Price
$15, one-time, no subscription. VAT included.

Related guide: [Building a multilingual AI content pipeline](https://forgehouse.ai/guides/ai-content-pipeline/)