How to use the Amazon Ads Audit agent skill (ads-amazon)

What is an agent skill, and what does ads-amazon make the agent do?

An agent skill is a folder containing a SKILL.md file that gives an AI agent specific instructions for a task. When the task matches the skill's description, the agent reads those instructions and follows them.

The ads-amazon skill tells the agent to review an Amazon Ads account in a defined order. It first gathers inputs such as your goal, conversion definition, account age, geography, date window, timezone, currency, spend and available data sources. It then checks the account against controls covering profiles and regions, measurement, portfolios, sponsored and DSP formats, targeting, search terms, retail readiness, creative, budgets, ACOS, TACOS and policy.

• Separates observations, diagnoses, recommendations, opportunities and proposed changes.
• Marks uncertainty and contradictions in the data.
• Returns findings as a structured report.
• Keeps any suggested account changes as drafts rather than applying them.

Install the ads-amazon skill

Download the ZIP for skills/ads-amazon from this page and unzip it. To make the skill available across all your projects, place the unzipped folder in ~/.claude/skills/. To make it available only inside one project, place it in that project's .claude/skills/ folder.

If you use the Claude apps, you can instead upload the skill under Settings → Capabilities → Skills.

• Personal install: ~/.claude/skills/ (applies to all projects).
• Project install: .claude/skills/ inside your project folder (applies to that project only).
• Claude apps: Settings → Capabilities → Skills, then upload.

Install the required reference files

The SKILL.md on its own is not fully self-contained. It refers to the main ads operating contract and to ads/references/amazon-audit.md, which come from the wider claude-ads toolkit and are not included in this file. The procedure says to resolve these from the plugin root (the directory holding ads/ and skills/) or from the skills directory in a standalone install.

If those reference files are missing, the audit will be less complete. Install the wider toolkit so the paths resolve before you rely on the results.

Trigger the skill with natural requests

You do not need a special command. Describe what you want in plain language, and mention Amazon Ads terms so the agent matches the skill. The skill description includes terms such as Amazon Ads, sponsored ads, Amazon PPC, ACOS, TACOS, ASIN advertising, Amazon DSP and retail-media optimization.

Give the agent the data you have. Exports, screenshots or API results work, as long as you say what they are and what period they cover.

• "Audit my Sponsored Products campaigns for the last 30 days. Target ACOS is 35%, currency is GBP."
• "Review my search term report and tell me where wasted spend is hiding."
• "Compare my ACOS and TACOS against my margin goal and list the issues before our budget meeting."
• "Check my Sponsored Brands and Sponsored Display setup for targeting and creative problems."

Provide the inputs the audit needs

The skill asks for key context before scoring. Without it, the agent will flag missing inputs rather than guess. Useful inputs include your objective, how you define a conversion, account and campaign age, geography, date window, timezone, currency, spend, targets and the data sources you can share.

Every number the agent uses is traced back to the export, screenshot, API result or manual value you supplied.

• Objective and conversion definition
• Account and campaign age
• Geography, date window and timezone
• Currency and spend
• Targets such as ACOS or TACOS goals
• Exports, screenshots or API results

Read the report and decide on changes

The report includes platform health, evidence coverage, regulatory exposure, observations, diagnoses, prioritised recommendations, unscored opportunities, contradictions, missing inputs and recovery hints. Treat each recommendation as a suggestion to check against your own data.

Any account change the agent proposes stays a draft. A person should review and approve edits before they are made in Amazon Ads.

• Check whether an observation is supported by the data you supplied.
• Read the contradictions and missing inputs section before acting.
• Approve or reject each draft change yourself.

Safety notes: scripts and network access

According to the skill's file listing, ads-amazon runs no scripts and uses no network access. The file describes no Amazon account credentials. The agent works only with the data you give it.

Because the skill asks the agent to treat external account and web content as data, not instructions, pasted text from reports should not be treated as commands. Benchmarks are not applied blindly; the skill requires checking them against objective, geography, methodology, sample size, conversion lag and account maturity.

• Runs scripts: no
• Uses network: no
• No Amazon account credentials are described in the file.