PluginPublic

kuudoai/kuudo-skills

22 Agent Skills for Amazon sellers and advertisers: Sponsored Products and Brands audits, search-term mining, budget pacing, DSP and Amazon Marketing Cloud, reporting, listing optimization and compliance, product images, repricing, FBA inbound, stockout, refunds, buyer messaging, and profitability. Built for Kuudo's hosted Amazon Ads and Selling Partner MCP servers; price, budget, listing, and shipment writes are previewed and wait for explicit approval.

1.65x

Quality

93%

Does it follow best practices?

Impact

66%

1.65x

Average score across 22 eval scenarios

Securityby

Medium

Suggest reviewing before use

Bullet Intent Scoring (COSMO framework)

Load this when auditing bullet quality for conversion or rewriting bullets. This is the optimization layer that sits on top of the hard compliance caps in 01-policy-rules.md § 3 — compliance keeps a listing live; intent coverage makes it convert. A bullet set can be 100% policy-compliant and still be a feature dump that answers no shopper question.

Why this exists

Amazon's ranking and its shopping assistant (Rufus) increasingly retrieve and summarize listings by shopper intent and context (the COSMO model), not just keyword match. Bullets that read as intent answers — who it's for, what job it does, where/when it fits, why to trust it — surface better and convert better than bullets that list specs with no connective tissue. The goal of a bullet is to answer a question a shopper is actually asking, then back it with evidence.

The four intent dimensions

Every strong bullet set covers all four across its 5 bullets. A single bullet can hit more than one. Use these as a coverage checklist, not a template.

Dimension	Answers	Signal vocabulary (anchors, not a whitelist)
Audience / need fit	Who is this for? What problem or preference does it solve?	"ideal for", "designed for", "for sensitive skin", "if you", "anyone who", "helps reduce", "relief from", "prevent", "support"
Function / use case	What job does it do? In what activity, event, or flow?	"use for", "use as", "clean / cook / store / carry / travel", "for camping / office / workout", "daily use", "on the go"
Context / compatibility	Where, when, or what does it work with?	"compatible with", "fits", "works with", "replacement for", "dishwasher safe", "indoor / outdoor", "waterproof", material/surface words
Decision evidence	Why trust it? What's the proof or differentiator?	specs and numbers, "includes / comes with", "set of", "certified / lab tested / BPA-free", "unlike", "patented", dimensions, counts, materials

Any bullet that contains a number (size, count, dimension, %) automatically carries some decision evidence — but a number alone is not a substitute for the other three dimensions.

Per-bullet scoring rubric (1–5)

Start each bullet at 5 and deduct. Floor at 1, cap at 5.

Condition	Deduct
Empty bullet	set to 1
Too short (< 50 chars) — no room to make a point	−2
Short (50–99 chars) — thin; ideal is 100+ for substance	−1
Too long (> 255 chars — also the hard policy cap)	−1
Vague marketing phrase ("premium quality", "high quality", "best in class", "world class", "industry leading", "revolutionary", "amazing", "incredible")	−1
Excessive ALL CAPS (> ~30% of the bullet, or 3+ caps words)	−1
Reads like a comma-separated feature list, not an intent answer (4+ short fragments, most ≤4 words)	−1
Addresses no clear shopper intent (none of the four dimensions)	−1
Carries intent but no concrete decision evidence (no spec/number/included-item/certification/material)	−1

These length thresholds (50 / 100 / 255) are substance heuristics for scoring. They do not override the hard policy rules in 01-policy-rules.md § 3 (10–255 characters, at least 3 bullets, capital start, sentence fragment with no end punctuation, header:description shape, numbers one–nine spelled out, unique content per bullet). Compliance gates first; intent score optimizes within.

The 255-character ceiling makes intent coverage harder, not optional — there is less room, so every clause has to earn its place. Cutting vague marketing ("premium quality", "best in class") is usually where the budget comes from.

Coverage rule (set-level)

After scoring the 5 bullets individually, check the set: do the five together cover all four dimensions? Flag any dimension missing across the entire set — that's a gap even if every individual bullet scored well. Most thin listings miss audience/need (they describe the product but never name who it's for) or decision evidence (claims with nothing behind them).

Tier labels (set average)

Avg score	Tier	Action
≥ 4.0	Good	Minor improvements possible
3.0–3.9	Fair	Several improvements needed
2.0–2.9	Weak	Major rewrite recommended
< 2.0	Critical	Bullets need a complete overhaul

How to apply

In an audit — Report a bullet-intent assessment: the set tier, which dimensions are covered vs. missing, and the 1–2 weakest bullets with the specific deduction reason. Keep it to the signal — don't print a five-row score table unless the user asks. This is the conversion half of the review; the compliance half (suppression, prohibited content) comes from 01-policy-rules.md.

In an edit — When rewriting bullets, make each one answer at least one dimension and ensure the set covers all four. Lead with the shopper outcome, then attach the evidence. Replace vague marketing with concrete attributes.

Worked example

Weak (score 2): PREMIUM QUALITY MATERIAL, durable, long lasting, great value, buy now Deductions: vague marketing (−1), excessive caps (−1), feature-list fragments (−1), no decision evidence (−1). Addresses no clear intent.

Strong (score 5): Built for daily commuters: the 18L water-resistant 600D polyester pack holds a 15.6" laptop plus a change of clothes, with a luggage pass-through that slides over a roller handle so it rides hands-free through the airport Covers audience/need (commuters), function (carry laptop + clothes), context (fits a 15.6" laptop, pairs with a roller bag, water-resistant), and decision evidence (18L, 600D polyester, dimensions). One sentence, one shopper, real proof.

Notes

These are universal optimization heuristics — they apply across categories. Category-specific bullet patterns live in the amazon_atlas knowledge base (per-category style guides, doc_type: style_guide); query that for category-specific phrasing conventions, then apply this intent rubric on top.
The vocabulary lists above are anchors to help you recognize intent signals. Judge in context — a bullet can answer "who it's for" without using any of the literal phrases. Don't reward keyword stuffing of these phrases; reward genuine intent answers.