Audit creator scripts for TikTok Shop and Amazon policy violations. Returns flags and safe rewrites.
BanProof AI MCP Server
BanProof AI is an MCP server that audits video scripts for TikTok Shop and Amazon policy violations. It scans scripts against TikTok Shop content policies, the Amazon Associates Operating Agreement, and FTC disclosure guidelines, then returns flagged phrases, violation categories, and safe rewrites designed to keep the hook and CTA intact.
🛠️ Key Features
Tool: audit_script
Inputs: script (full video script text)
Outputs: flagged phrases, violation categories, safe rewrites, overall r
🚀 Use Cases
Pre-publishing review of creator scripts for TikTok Shop affiliate workflows
Pre-record audits of Amazon affiliate video scripts
Checking FTC disclosure guideline compliance within video script text
⚡ Developer Benefits
Pinpoints exact phrases that trigger violation points or account strikes
Provides safe rewrites for policy-compliant alternatives while preserving hook and CTA
⚠️ Limitations
Scope is limited to TikTok Shop content policies, Amazon Associates Operating Agreement, and FTC disclosure guidelines
The server description does not specify support beyond the audit_script tool
Audit TikTok Shop and Amazon affiliate video scripts for policy violations before creators record.
What it does
BanProof AI scans video scripts against TikTok Shop content policies, Amazon Associates Operating Agreement, and FTC disclosure guidelines. It flags the exact phrases that trigger account strikes, returns safe rewrites, gives each script a BanProof risk score (our own severity rating, not a platform score), and generates ready-to-submit appeal text when you've already been hit.
Tools
audit_script
Scan a script for policy violations before recording.
Input:
script (string, required) — full video script text
product_url (string, optional) — product page URL to cross-check script claims against actual product claims
Output: flagged phrases with reasons and safe rewrites, overall risk level (safe / elevated / high), and a BanProof risk score per violation and in aggregate
Detects:
Medical / health claims (cures, clinically proven, FDA approved, before/after without disclaimer)
Guarantee and refund guarantee claims
False certifications and unverified endorsements
Urgency / scarcity language (limited time only, only 3 left, countdown timers)
Fake social proof (going viral, everyone is buying, unverified sales volume)
Income claims (earn $500/day, passive income stream)
Missing FTC disclosure (#ad / #sponsored absent on commercial content)
Price claim violations (lowest price ever, was $X, save X% without source)
AIGC / AI voice disclosure requirement (TikTok policy 2024, AI voice LIVE ban July 2026)
Third-party testimonials and before/after transformations without required disclaimers
Product overclaims vs the actual product page (when product_url provided)
generate_appeal
Write a ready-to-submit appeal for a platform violation notice.
Input:
violation_notice (string) — the full text of the violation notice you received
platform (tiktok | amazon | youtube)
Output: appeal text ready to paste into the platform's appeal form, with character count vs platform limit (TikTok: 800 chars)
Connect
Hosted endpoint:https://banproof.io/mcp
Auth: OAuth 2.0. Create a free account at banproof.io — 3 credits/month, no credit card required.
Compatible with Claude, Cursor, and any MCP-compatible client.
Example — audit_script
Input:
code
This serum will completely cure your acne in 3 days — guaranteed results or your money back. FDA approved and clinically proven. Only 2 left in stock, limited time only! Link in bio.
Output:
code
⚠ 5 violations detected. Risk level: high — BanProof risk score 7/12
1. [Absolute medical claim · 2pt] "cure your acne in 3 days" → "support clearer-looking skin over time"
2. [Guarantee / results claim · 1pt] "guaranteed results or your money back" → "so many creators love how it performs"
3. [False certification claim · 2pt] "FDA approved" → "made with carefully selected ingredients"
4. [False scarcity claim · 1pt] "Only 2 left in stock" → "while supplies last"
5. [FTC disclosure required · 1pt] "Link in bio" → (add #ad or #sponsored)
Pricing
Plans are metered in credits: script audit = 1, video scan = 3, appeal = 2.