18 policy rules · 15 free tools & checkers
Amazon & Shopify copy, compliance-scanned before you publish
CopyForge generates listing, ad, SEO and email copy from the facts you supply — then scans every line against the applicable rules in its 18-rule library and flags the exact phrase, with its dated policy source, before you publish. Future-dated rules activate on their effective date.
No card required for the free plan. Start with 10 generations a day, on every channel.
Anything can write a paragraph. The scan is the product.
Every line the generator produces — and every line you paste — runs the same gauntlet before it goes near your listing.
Flags with receipts
Every flag names the exact phrase it tripped and cites the dated marketplace policy behind the rule — so you can verify the call, not just obey it.
18 rules, versioned
The ruleset ships as a versioned engine (currently 2026.09.2) and is re-reviewed on a dated cadence — not a vague keyword filter.
Channel hard limits
Title lengths, bullet counts and field structures are shaped to configured marketplace limits before export, reducing avoidable format rejections.
Diversion detection
Emails, phone numbers, bare domains and links are checked against the configured diversion patterns so common off-platform contact risks can be reviewed before publication.
No card required
The free plan stays free
10 a day free
Every channel included
No CopyForge model training
Your generations stay account-scoped
Cancel anytime
No lock-in, no contracts
Ruleset v2026.09.2 · rules last reviewed 23 September 2026 · every flag quotes the exact phrase and links its dated source · 15 free tools & checkers · 21 seller guides · no card to start
Five channels, 15 free tools & checkers
A good Amazon title is a bad Google ad. Five generators write to each channel’s own field structure and hard limits — while the rest of the public toolbox provides dedicated title, keyword, claim, policy and store checks. No account is needed to try the public tools: previews and checkers run for free, and signup adds saved history and a higher daily generation quota.
Amazon listing
Title under 75 characters, Item Highlights within 125, exactly five bullets, backend keywords under 250 bytes (249 safe max).
Shopify description
Product description plus a meta title under 60 characters and meta description under 155.
SEO / AEO / GEO
SEO metadata plus optional AEO direct-answer and GEO entity-summary modes, grounded in your supplied facts.
Facebook / social ad
Three variants under 125 characters, screened against ad-platform policy language.
Email campaign
Four subject lines under 50 characters and a body written to CAN-SPAM norms.
All 15 public tools and checkers are free to use — open the full toolbox.
How it works
CopyForge connects product facts, supporting evidence and approved claims with channel rules — so you know not only what wording may be risky, but what supports it, where it may be used, and when it needs review again.
Give it the facts
Paste your specifications, materials, dimensions and warranty terms. They are saved as structured product records, and the model is instructed to treat this block as the complete set of permissible factual content.
Attach evidence, approve claims
Upload supporting documents to the private evidence vault, register the claim wording they support and route it for approval — so a claim is only reusable once someone has signed off on it.
Generate for a channel
Each channel has its own field targets and character limits baked into the prompt, so output arrives already shaped for where it is going.
Scan, review, monitor
Every output is scanned against the applicable rules in the 18-rule library, with the flagged phrase, rule and reason. Future-dated rules activate on their effective date. Import catalog listings for on-demand scans, and daily monitoring flags content, rule or evidence changes that need review again.
Prove it on your own listing — right now
The fastest trust test is your own copy. Paste a product title or bullet points below and run the same rule scan the generator applies to every output — here on the page, before you create an account.
Free — no signup, nothing saved. Up to 5 checks a day per connection.
The guardrails are the product
Anything can generate a paragraph. The useful work is surfacing common claim patterns and format problems before you decide what to publish.
Facts-first specification guardrail
The prompt forbids stating any number, material, certification or rating that is not in your input; the result still requires human verification.
Brand claim allowlist
Pro brand profiles restrict the model to claims you have marked as evidenced, and reject your banned words.
Automatic repair pass
If a forbidden term slips through, the violation is fed back and only the affected passages are rewritten.
Your account data stays scoped
Authenticated data queries use row-level security and explicit ownership filters. CopyForge does not train its own models on your generations.
Want to inspect the scanner instead of trusting a score? Read all 18 checks and their policy sources.
See exactly what you get
Paste your product facts, get channel-formatted copy with a compliance score and the flagged phrases spelled out — before you copy anything to a listing.
Your facts
One rule matched — review before publishing
The exact phrase and source are shown below.
1.7L Electric Kettle, 2200W — Auto Shut-Off
World-class performance from a 2200W element · 304 stainless steel interior · auto shut-off when dry
Unsupported superlative
Matched: “World-class” — replace it with a measurable product fact.
Repair pass — 100 score, no configured issues detected
“World-class” rewritten to a fact you supplied: “2200W concealed element”. The rewrite is re-scanned before it reaches you — configured checks are not marketplace, legal or factual approval, so verify details before publishing.
Inside the dashboard
Real screenshots of the product, not mockups: the usage overview, the channel brief that grounds every generation, and the compliance scan that follows it.
Why not just use a general AI chatbot?
A chatbot can draft a paragraph. CopyForge adds a structured facts-first workflow and a deterministic policy pass so marketplace copy is easier to review consistently.
| Capability | CopyForge | General AI chatbot | Writing by hand |
|---|---|---|---|
| Grounded in the facts you supply | Facts-first prompt with a clear instruction not to fill gaps; human verification still required | Optional — the model may fill gaps unless you constrain it | Yes, if the writer stays within the supplied facts |
| Policy check before you publish | Applicable checks from a 18-rule documented library, with the exact flagged phrase | Only when prompted, without CopyForge’s deterministic rule pass | Only the checks the writer remembers to perform |
| Consistent brand voice and banned words | Brand profiles apply approved claims and forbidden words every run | Guidelines normally need to be supplied in the conversation | Depends on the writer and review process |
| Current public workflow | Single-item generation on Free; available paid plans add bulk and exports | Flexible conversation, with output checked and organized manually | Every draft and check is completed manually |
| Cost | Free plan: 10 generations per day; available paid plans start at $19/month | Varies by provider and plan | The writer’s or copywriter’s time |
This compares approaches, not named products. Provider capabilities and prices vary; the CopyForge column distinguishes live Free access from available paid plans.
Frequently asked questions
- What happens to my product data?
- Generations stay scoped to your account, and CopyForge does not use them to train a model. You can export or delete your data from the dashboard at any time.
- Which channels and marketplaces are covered?
- Amazon listings, Shopify product descriptions, SEO/AEO/GEO search-discovery content, Facebook and social ads, and email campaigns — each with its own field structure and policy checks. Etsy- and eBay-specific rule sets are on the roadmap.
- Which languages does it write?
- English output today, because the policy rules it checks are written for English-language marketplaces. More languages are planned.
- Where does the attribution line appear?
- On the free plan, generated text ends with "Generated with CopyForge AI — https://copy-forge-ai.com". It is part of the text you copy out. Pro and Agency remove it.
- Do I need a card to start?
- No. The free plan needs no card, gives you 10 generations a day on every channel, and never converts to a paid plan automatically.
Start free. Scale only when the workflow earns it.
The free plan gives you 10 generations a day and one brand profile — enough to judge the workflow on your own catalogue. Pro and Agency limits are published on Pricing — every cap is shown before you pay, and there are no surprise usage charges.
Why use a purpose-built workflow? A general chat model can produce unsupported claim wording when the prompt leaves room for it. CopyForge constrains generation to the facts you supply and then runs a deterministic scan that reports configured flags with the phrase, rule and dated source. Human verification still matters before publication.
Free — $0 forever, 10 generations a day, no card · Pro — $19/mo · Agency — $49/mo. Every cap is published up front on Pricing.