CtrlK
BlogDocsLog inGet started
Tessl Logo
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.

68

1.65x
Quality

93%

Does it follow best practices?

Impact

66%

1.65x

Average score across 22 eval scenarios

SecuritybySnyk

Medium

Suggest reviewing before use

Overview
Quality
Evals
Security
Files

report-template.mdskills/amazon-ads-sp-campaign-auditor/references/

Sponsored Products campaign portfolio audit

Fill requested sections from the validated Amazon Ads MCP report. Mark unsupported sections N/A and name the missing field or unresolved ambiguity.

1. Scope and provenance

  • Identity, advertiser account/profile, region, marketplace, and currency
  • Date range and report ID
  • Report-field catalog or schema version when exposed
  • Exact requested and accepted field IDs
  • Returned row count and detected grain
  • Aggregation and budget-denominator rules

2. Executive findings

Summarize the most material observations. Label each statement as:

  • Observation: directly calculated or returned by the MCP
  • Hypothesis: a plausible explanation requiring investigation
  • Recommendation: a reversible next step tied to evidence

3. Portfolio composition

  • Unique campaigns after aggregation
  • Delivery-status counts, with proxy status disclosed
  • Spend and optional attributed sales totals
  • Targeting-type mix when supported by a validated field
  • Naming-pattern counts only when the user confirms the taxonomy

4. Spend concentration

  • Top-N campaigns by cost, including campaign ID, name, cost, and portfolio share
  • Total cost and top-N share formula
  • User-provided threshold and result, or descriptive distribution when no threshold exists

5. Budget fill

  • Campaign or portfolio fill rates only where window-aligned denominators are available
  • Budget type, eligible-day rule, and treatment of budget changes
  • N/A explanation where historical budget cannot be reconstructed

6. Delivery and engagement review

  • No-delivery campaigns
  • Visible-without-engagement campaigns
  • Data-quality anomalies
  • Prioritized review table with evidence and diagnostic hypotheses

7. Structure and naming

  • Observed targeting mix
  • Confirmed naming taxonomy analysis
  • Normalized duplicate-name groups, retaining campaign IDs

8. Recommendations

Provide a short, evidence-linked review queue. State what should be verified before any change and identify risks such as seasonality, learning effects, shared constraints, or recent launches. Do not execute campaign changes.

9. Limitations

List missing fields, proxy measures, unresolved account context, schema compromises, threshold sources, and any section marked N/A.

skills

LICENSE.md

README.md

tile.json