| name | listing-optimizer |
| description | Generate or rewrite Amazon product listing content: titles, bullet points, product descriptions, and backend search terms. Use when the user wants to create a new listing, rewrite existing copy, optimize specific elements, or generate Amazon-compliant product content. Handles full listings, individual elements (e.g., "just fix my title"), and variation-child copy differentiation when multiple products inherit shared parent positioning.
|
Listing Optimizer
Creates Amazon-compliant, conversion-optimized listing content that balances
traditional A9/A10 keyword search with Rufus AI conversational discovery.
Prerequisites
Before generating any content, you MUST have:
- Product information — What is it? Materials, features, specifications,
dimensions, what's included.
- Target audience — Who buys this? What problem does it solve?
- Differentiators — What makes it different from competitors?
- Category — Which Amazon category? (Affects character limits)
- Keywords — Primary and secondary target keywords
(If not provided, generate suggestions and confirm with the user)
- Marketplace — US, UK, CA, etc. (Affects backend byte limits)
Load references/amazon-style-guide.md for current rules.
Load references/rufus-optimization.md for AI discovery optimization.
Load references/backend-keywords-guide.md for backend string construction rules.
If brand-context.md exists in the project root or .agents/ directory,
load it for shared brand voice and past learnings.
If products/[ASIN]/product-context.md exists, load it for product-specific
facts. Prefer User Confirmed Data, then Amazon Observed Data, then AI Inferences.
Resolve product references by SKU or ASIN using the product context file's
SKU: and ASIN: fields.
If multiple product folders exist and the user has not specified a product, ask
which SKU or ASIN to use before writing.
If the user wants to differentiate child ASINs within one variation family,
load the parent product context plus every relevant child product context before
writing. Use confirmed variation fields such as Parent ASIN, Variation family, set_name, shape, count, and capacity when available.
Never assume shared mechanics across variants. If a construction detail, seal
mechanism, included accessory, or capacity claim is not confirmed in product
data or by the user, treat it as unknown and ask or leave it out.
Workflow
Step 1: Interview (if needed)
If the user hasn't provided sufficient product info, interview them using
the framework from templates/listing-brief.md. Keep it conversational.
Minimum viable info to start: product name, 3-5 key features, target buyer.
If this is a variation-child differentiation request, confirm only the missing
child-specific facts needed to safely customize copy:
- Which parent ASIN or parent product context governs the family
- Which child ASINs or SKUs should be rewritten
- Which child attributes are confirmed:
set_name, shape, count, capacity
- Whether materials and seal mechanism are truly identical across the family
- Any child-specific positioning: entry set, combo, flagship, travel size, etc.
Step 1A: Variation Child Differentiation Mode
Use this mode when the user wants to customize copy across variation children
that share inherited parent content.
Required Inputs:
- Parent
product-context.md
- All relevant child
product-context.md files
- Confirmed child attributes, especially
set_name, shape, count, and
capacity
- Any confirmed notes on what is shared across the family vs. unique per child
Differentiation Rules:
- Keep the family-level strategy aligned so the variation set still feels like
one product line
- Make Bullet 4 fully unique for each child: name the shape, count, and one
shape-specific or configuration-specific benefit that is actually supported by
the product data
- Adapt Bullets 1 and 5 so the use-case examples match the child's natural fit
and positioning
- Keep Bullets 2 and 3 identical only when the physical construction,
materials, and seal mechanism are confirmed to be the same across the family
- Rewrite the description to remove inherited parent references and reposition
the value proposition for that specific child
- Add child-only backend terms that match the actual shape/configuration and
remove terms that do not apply to that child
Use-Case Guidance:
- Rectangular variants naturally support casserole, leftovers, or fridge-shelf use
- Square variants naturally support stacking, pantry organization, or shelf efficiency
- Combo variants naturally support flexibility, mixed meal prep, or households
that want multiple formats
Treat the examples above as directional only. Use them only if they match the
confirmed variant data.
Step 2: Generate Title
Process:
- Identify the primary keyword (highest search volume, most relevant)
- Check the category character limit (200 default, lower for some categories)
- Write for mobile first: critical info within first 70-80 characters
- Apply the category title formula from the style guide
- Verify compliance: no prohibited characters, no word repeated more than twice,
no promotional language
Title Structure:
[Brand] [Primary Keyword Phrase], [Key Feature/Benefit], [Secondary Feature] - [Size/Count/Variant]
Self-Check Before Presenting:
Step 3: Generate Bullet Points
Process:
- Map out 5 distinct selling angles — no overlap between bullets
- For each bullet, write in Header: Body format
- Connect every feature to a real-world benefit and use case
- Integrate 2-3 keywords per bullet naturally (never forced)
- Keep each bullet under 200 characters for safe indexing (total under 1,000 bytes)
- Front-load the strongest bullet (most visible, especially mobile)
5-Bullet Framework:
| Bullet | Focus | Pattern |
|---|
| 1 | Primary benefit / use case | What it DOES for the buyer + when/where/how |
| 2 | Quality / materials | What it's MADE of + why that matters |
| 3 | Convenience / ease of use | How EASY it is to use/maintain + compatibility |
| 4 | What's included / specs | What you GET + specific numbers/dimensions |
| 5 | Who it's for / trust signal | WHO benefits + social proof or brand credibility |
Each Bullet Structure:
[Benefit Header]: [Feature detail that proves the benefit, with specific
numbers or materials, connected to a real use case the buyer recognizes]
Example (for a glass vacuum container):
Keeps Food Fresh Up to 5x Longer: One-touch vacuum pump removes air to slow
oxidation and bacterial growth, keeping berries, cut vegetables, and leftovers
fresh for days instead of hours
Premium Borosilicate Glass: Unlike plastic containers, this lab-grade glass
will not stain, absorb odors, or leach chemicals — safe for microwave,
freezer, and dishwasher without warping
Simple One-Hand Operation: Press the silicone valve on the lid to seal in
seconds — no separate pump needed, no batteries, no complicated parts to lose
Complete Starter Set: Includes three nesting containers (12 oz, 24 oz, 34 oz)
with interchangeable lids, a silicone sealing ring replacement, and a quick-
start guide
Designed for Meal Prep and Busy Families: Stackable space-saving design fits
neatly in standard refrigerators, with measurement markings for portion control
and grab-and-go convenience
Self-Check Before Presenting:
If differentiating variation children, apply these bullet rules:
Step 4: Generate Product Description
Process:
- DO NOT repeat bullet points — expand, don't duplicate
- Use the 5-step framework: Problem → Solution → Specifics → Objections → Confidence
- Keep under 2,000 characters
- Include secondary and long-tail keywords not covered elsewhere
- Short paragraphs (2-3 sentences max for mobile readability)
- If the seller is Brand Registered, note that A+ Content should ALSO be created
(description feeds the algorithm; A+ Content converts shoppers)
Framework:
[Open with the buyer's problem or desire — what brought them to search]
[Present your product as the solution — connect to their specific situation]
[Support with specifics — materials, dimensions, certifications, testing]
[Handle the top 1-2 objections buyers have in this category]
[Close with a confidence statement — what using this product feels like,
or a practical next step]
Self-Check:
Step 5: Generate Backend Search Terms
Process:
- List all unique words already in title, bullets, and description
- Identify gaps: synonyms, Spanish translations, use-case terms, misspellings
- Build the string per
references/backend-keywords-guide.md
- Count bytes — must be under 249 (target 235 for safety buffer)
- Separate with single spaces, no commas, no punctuation
Present the backend string AND the byte count to the user.
Self-Check:
Output Format
Present the complete listing in this order:
## Product Title
[title text]
Characters: [count] / [limit for category]
## Bullet Points
1. [bullet 1]
2. [bullet 2]
3. [bullet 3]
4. [bullet 4]
5. [bullet 5]
Total characters: [count] | Estimated bytes: [count] / 1,000 indexing limit
## Product Description
[description text]
Characters: [count] / 2,000
## Backend Search Terms
[keyword string]
Bytes: [count] / 249
## Compliance Notes
[Any category-specific warnings or recommendations]
## Rufus Optimization Notes
[Which shopper questions this listing can now answer]
After presenting, ask the user:
- "Does this match your brand voice?"
- "Any features or use cases I should emphasize more?"
- "Want me to audit the keyword coverage or adjust any section?"
Variation Child Output Mode
When the user asks for differentiation across a parent/child variation family,
write the proposed copy into each child's products/[ASIN]/product-context.md
under ## Listing Draft.
Use this structure:
## Listing Draft
### Bullet Drafts
1. [bullet 1]
2. [bullet 2]
3. [bullet 3]
4. [bullet 4]
5. [bullet 5]
### Description Draft
[description text]
### Backend Search Terms Draft
[keyword string]
### Differentiation Notes
- Shared bullets kept from family: [bullet numbers or "none"]
- Child-specific angles introduced: [short notes]
- Facts still requiring confirmation: [short notes or "none"]
If the user wants drafts only in chat, present the same per-child structure in
the response instead of editing files.