Strip the slop. Keep the insight. Automatic quality control for AI prose.

Connect a high-performance quality gate directly to your AI agents (Claude Code, Claude Desktop, ChatGPT, Suits.ai, and Gemini CLI). Run smart editorial checks instantly on our secure global servers—filtering robotic cadences, overused buzzwords, and try-hard style.

Create account How to use

MCP Tools

Your quality toolbox. Six tools your agent can call right now.

Every tool runs on our secure global servers via the Model Context Protocol. Use them individually for surgical precision — or let the Sage orchestrate the entire pipeline for you.

The Sage Primary tool

writing_quality_sage

The mastermind. One call checks your content for substance, AI patterns, hokiness, and AP Style — then automatically rewrites it until every pass reports CLEAN.

1
Substance Flags vague, content-free language. Rewrites for concreteness using a cold-reader sub-agent.
2
Guard + Hokiness Scans 25+ deterministic AI tells and filters try-hard, grandiose tone. Rewrites combined violations.
3
AP Style Deterministic regex pass plus LLM judgment for AP Stylebook compliance. Rewrites violations.

Iterates up to 3 rounds per phase (or 15 in perfect mode). Content hash deduplication prevents infinite loops.

quality_check

The primary detection-only tool. Runs all four engines — AI tells, substance, hokiness, AP Style — and returns a unified violation report. No rewrites; use this when you want to see the raw findings. Also available as writing_guard_check.

Detection engine used by the Sage

substance_check

Standalone LLM-based evaluation for meaning, consistency, and zero-information patterns. Quick audit of how concrete and specific a draft actually is.

Included in the Sage · Phase 1

hokiness_check

Identifies try-hard, grandiose, or "motivational poster" language. Checks whether text reads like a real person or AI-speak. Supports social and editorial modes.

Included in the Sage · Phase 2

ap_style_check

Verifies compliance with AP Stylebook rules across two passes — deterministic regex for clear-cut violations, then LLM judgment for nuanced style calls.

Included in the Sage · Phase 3

get_skill_procedure

Returns the AI Writing Guard skill overview and usage instructions. Use this when your agent needs to review the recommended workflow for integrating the quality-check tools.

Utility tool

Proven Results

Proven across thousands of client engagements.

The Guard is grounded in a pattern library built from InnovAItion Partners' real-world client work. We've cataloged the specific fingerprints — from em-dash overuse to suppressed truth framing — that flatten otherwise useful LLM output.

25+ Style Detectors mapping to structural AI tells
Millisecond Speeds on secure global servers
Thousands of Drafts refined using real-world editorial insight

Privacy & Security

Your data stays yours. We just provide the guardrails.

AI Writing Guard is designed with a privacy-first architecture. We focus on the patterns, not the payload.

Zero Retention

We do not store any of the content you send through the guard. Once the analysis is complete, the text is purged from our volatile memory immediately.

No Model Training

We never train models on your content. The core detection engine runs deterministically via compiled code. When using optional rewrite tools, your content is handled temporarily under strict volatile edge memory constraints.

Secure & Reliable

Our systems are built on enterprise-grade infrastructure, ensuring your data is handled with the highest security standards throughout its brief transit.

Minimal Logging

We only store usage statistics (metadata) for tracking and billing purposes. We never log user-supplied content or the specific results of your guard runs.

Centralized Quality

Why a Centralized Guard beats manual prompting.

Prompt guidelines drift, change, and fail between models. A centralized, secure quality gate guarantees consistent standard enforcement across all teams.

Benefit
Centralized Writing Guard
Manual Writing Prompts
Consistent Enforcement
One consistent quality standard automatically shared across all team members and tools.
Subject to model interpretation drift and prompt decay.
Secure Team Accounts
Secure team access keys & private usage reports to easily track use and brand alignment.
Zero central visibility or compliance tracking.
Extreme Speed
Instant checks powered by secure global servers.
Bloats LLM context windows, slowing and inflating token cost.
Composable Tooling
Plug-and-play integration with Claude Desktop, Suits.ai, and your team's favorite apps.
Must be manually copy-pasted and adapted for each workspace.

Tooling

Connect the guard to your existing LLM toolchain.

AI Writing Guard integrates directly with Claude Desktop, IDEs, and custom agents via the Model Context Protocol (MCP). Learn how to connect the guard in minutes.

View the full guide 
{
  "mcpServers": {
    "writing-guard": {
      "url": "https://ai-writing-guard.mcp.doorw.ai/mcp",
      "headers": {
        "Authorization": "Bearer YOUR-API-KEY"
      }
    }
  }
}
Looking for developer tools? View Python CLI & Agent integration details
$ aiwg check draft.md --type blog
❌ FAILED - 3 violations.
HYPOPHORA (1): Short rhetorical question...
BUZZWORD (1): Banned AI buzzword: "delve"
AP_STYLE (1): Capitals/numbers compliance...
$ aiwg rewrite draft.md --provider remote
Job initiated. ID: 4f2a... Polling for results...
Iteration 1 complete. Content is now CLEAN.

CLI Companion & Agent Skills

Universal alignment for humans and agents alike.

Integrate the Guard seamlessly into developer environments and publishing workflows. Use the companion Python CLI for local scripts, CI/CD audits, and text editors, or distribute our canonical SKILL.md system prompt.

  • Agent Synchronization: Direct your AI assistant to check, ingest decisions, and verify its own writing using a standardized MCP procedure.
  • CI/CD Linting: Fail builds when automated scripts detect robotic buzzwords or structural Tells in raw documentation.

Pricing

Start small. Scale your quality gate as your content volume grows.

Choose the capacity that fits your workflow. From trial-grade testing to high-volume team output, our tiers ensure consistent, human-sounding results at any scale.

Try it

Free

$0

Perfect for writers testing the quality checks on occasional drafts.

Words per day
5,000 words / day
Rate limit
3 requests / minute
Max input
500 words per call
  • Core style check alerts
  • Instant writing feedback
  • Standard writing formats
Create account - Free
Most practical

Teams & creators

Pro

$99/mo

$29/mo

First year — new users

Perfect for active content creators and professional editors checking drafts daily.

Words per day
25,000 words / day
Rate limit
20 requests / minute
Max input
2,500 words per call
  • Unified editorial pipeline checks
  • AP Stylebook compliance checking
  • Substance & Hokiness audit checks
  • Automatic memory for approved words
  • Built-in rewrite assistant
Continue to checkout

Org-wide

Enterprise

Custom

Private team accounts, detailed usage reports, and dedicated voice setup support.

Words per day
50,000 words / day
Rate limit
100+ requests / minute
Max input
10,000 words per call
  • Unified editorial pipeline checks
  • AP Stylebook compliance checking
  • Substance & Hokiness audit checks
  • Unlimited daily checks
  • Private team usage reports
  • Custom brand voice profiles
Contact sales