ads-amazon — independently scanned and version-tracked by SaferSkills.
SaferSkills independently audited ads-amazon (Agent Skill) and scored it 100/100 (green). The audit ran 55 deterministic rules across Security, Supply Chain, Maintenance, Transparency, and Community; it found 0 high-severity and 0 lower-severity findings. The full rule-by-rule trace and per-finding evidence are below. Free, methodology-open.
Findings & checks · 0 flagged
Every scanned point with the score it earned and what moved between them.
First recorded scan — no prior version to compare against.
The primary manifest — the file an agent reads to learn what this artifact does.
Amazon owns ~80% of US retail media (~$56B in 2025). This sub-skill audits the three Sponsored ad types plus basic DSP visibility for sellers and vendors. Detailed Amazon DSP audit (programmatic / Twitch / Fire TV / Freevee / Prime Video) is queued for Wave 3 in ads-retail-media.
Note: This is a leaf sub-skill. The thresholds and rules below live inline. In Wave 3, onceads-walmartships and theads-retail-mediaorchestrator extracts the shared retail-media logic, a dedicatedretail-media-specs.mdreference file will replace these inline rules (planned, not yet created — do not link to it from other files).
Campaign Performance Report, Targeting Report, Placement Report, Bulk Operations file
Search Terms, Repeat Purchase Behavior, Item Comparison
seller strategy on portfolio, coupons, and Sponsored Brands video)
ads/references/benchmarks.md for Amazon-specific benchmarksads/references/scoring-system.md for weighted scoringConsideration / Conversion) or by product line, not ad-hoc
SP_Brand_Exact_USD_Conv / SB_NonBrand_HSA_Video_USD_Cons)
cluster for search-term harvesting (auto-mining workflow)
(Exact, Phrase, or Broad) AND ≥1 Auto campaign
targeting) + Auto (for harvesting). Pure Auto is a learning-only state
brand storefront; SB Video tested for high-AOV products
Contextual (product / category targeting) campaigns — don't mix in one
search terms from Auto, promote winners to Manual Exact in their target campaign, add as negatives to the Auto so spend redirects
ASINs as negatives (prevents Auto cannibalizing Manual placements)
(prevents Sponsored Display contextual targeting from showing your ad next to incompatible products)
conversion terms above $10 / 30 days should be negatives
campaign with exact-match brand keywords and competitor-name negatives
if competitor bid is above your brand-defense bid
ACOS = (1 - contribution_margin) × buffer). Sloppy "30% across the board" ACOS targets cap performance
total ad spend / total revenue including organic; downward trend means paid is lifting organic, not replacing it
capture lift; auto-flag campaigns that need temporary ACOS uplift
hours differ from purchase hours (frequent for high-AOV electronics)
Up and Down for Manual Exact convertingcampaigns; Down Only for Auto campaigns or Sponsored Display contextual (avoid bid escalation in learning campaigns)
ROAS by placement (Placement Report); often +50-100% for top-of-search, -30% for product page placements
spend (caps strangle algorithm learning)
Operations not ad-hoc in the UI
page (Store, custom URL, or product list)
Required for high-AOV products
CVR than product detail page)
want top-of-funnel competitor traffic; remove if you do
Sell, and Lookalike audiences each scaled with appropriate bids
(your own complementary products + competitor ASINs you can win)
custom creative tested for top SKUs
separately — different CPC dynamics, different attribution windows
Item Comparison, Repeat Purchase Behavior
competitors on category keywords
strategy
ads driving to Amazon listings
| Metric | Pass | Warning | Fail |
|---|---|---|---|
| ACOS (vs target) | Within ±10% of target | 10-25% over | >25% over |
| TACOS trend (90d) | Decreasing | Flat | Increasing |
| Auto / Manual mix | Both active per ASIN | Only Manual or only Auto | Single campaign / ASIN |
| Negative keywords on Auto | Manual ASIN keywords added | Some missing | None |
| Search term harvest cadence | Weekly | Monthly | Ad-hoc |
| Branded campaign ACOS | <5% | 5-15% | >15% |
| Dynamic bidding strategy | Up-and-Down on converters, Down-Only on learners | Mixed inconsistently | All campaigns same strategy |
| Sponsored Brands video tested | Active on top 10 ASINs | Active on 1-3 ASINs | Not tested |
| Brand Analytics accessed | Weekly | Monthly | Never |
Amazon Ads Health Score: XX/100 (Grade: X)
Campaign Structure: XX/100 ████████░░ (15%)
Search-Term Harvesting: XX/100 ██████████ (25%)
ACOS / TACOS Discipline:XX/100 █████████░ (20%)
Bid & Budget Mgmt: XX/100 ████████░░ (15%)
Sponsored Brands: XX/100 ███████░░░ (10%)
Sponsored Display: XX/100 ███████░░░ (10%)
Brand Analytics: XX/100 █████░░░░░ (5%)AMAZON-ADS-REPORT.md: Full per-category findingsadd as negatives, ranked by spend recovery $
and audience size)
~30 seconds. Free. No account. Every finding cites a rule and a line of evidence.