Four Claude Skills That Audit Your Amazon Account Like an Analyst Would

What is inside

Four complete analysis systems, not four prompts

Each skill is a full method: required inputs, account-relative thresholds, a fixed report structure, and rules that force the reasoning into the open. Install them once and they trigger whenever you bring the matching task.

PPC Audit

Every search term classified through three lenses: wasted spend, structure debt, blind spots. Negatives list and a bulk-upload harvest table included. Conservative by design; thin data goes to a watch list, never to the chopping block.

You bring: your Search Term Report export

Listing Diagnostics

A scored teardown of your title, bullets, description, and backend terms, with compliance risk weighted as heavily as your title. Every flag quotes the exact phrase and shows the compliant direction.

You bring: your listing content, pasted

Keyword Evaluation

Turns any keyword export into a ranked decision: track and target, harvest later, conquest only, or prune with reasons. Works with exports from tools you no longer pay for.

You bring: any keyword CSV, from any tool

Competitor Teardown

Review mining, expectation gaps, and a gap map that ends in a differentiation decision, built entirely from listing text and reviews you collect. Honest sample counts on every claim.

You bring: a competitor listing and 20+ reviews

Plus the Prompt Pack. Ten standalone prompts: negative keyword triage, review mining, unit economics, event readiness, product validation, and more. Fill the brackets, paste, run.

The method

Every account problem falls into one of three lenses

The flagship PPC Audit organizes every finding the way a senior analyst would.

Wasted spend

Targets that spend and do not convert, week after week, kept alive because turning things off is nobody's job.

Structure debt

Accounts organized for the convenience of whoever manages them, not for performance. You cannot optimize what you cannot read.

Blind spots

The work that was never on anyone's checklist. Winners starved of budget while losers feast.

How it works

Export. Run. Approve and apply yourself.

1

Export your data

Pull the report from your own console: a Search Term Report, a keyword CSV, or pasted listing text. Your data never leaves your Claude conversation.

2

Run the skill

Attach the file and ask for the audit. The skill enforces the method: account-relative thresholds, fixed report structure, reasoning shown on every finding.

3

You approve every change

Every output ends at a recommendation with its evidence. You review it, you apply it by hand, and nothing touches your account without you.

Read this before you download

What the toolkit does, and exactly where it stops

The toolkit (free, no account access)

Audits, teardowns, diagnostics on data you export yourself
Shows its reasoning on every single recommendation
Needs no credential, no connection, no permission
Ends at a decision you apply by hand

What a one-shot run cannot do

Watch your account overnight and every night
Propose changes and wait in an approval queue
Verify results with follow-up reports after you approve
Run the same audit again next week, without you

THE TOOLKIT STOPS HERE

Everything below the line is what the Sydon agents do: the same analysis, running continuously, proposing changes with the reasoning shown and waiting for your approval before anything touches your account. If you find yourself re-running these skills every Monday by hand, that is the signal. See the platform.

Questions

Asked before downloading

Does this connect to my Amazon account?

No, and it cannot. The toolkit contains no integration, requests no credential, and explicitly instructs Claude never to ask for account access. You export your own reports and paste your own content. That is the entire data path.

Which Claude plan do I need?

Any plan that supports file attachments works for the prompts and the Projects format. Installing the packaged Skills requires a plan with Skills support. Every skill also ships as copy-paste Project instructions, so no one is locked out.

Is this just a bundle of prompts?

The Prompt Pack is ten prompts. The four Skills are more: each one enforces a complete method with required inputs, account-relative thresholds, a fixed report structure, and rules such as never recommending a pause on statistically thin data. The method is the product.

Will the recommendations be safe to apply?

Every recommendation ships with the numbers that produced it, and the skills are deliberately conservative: borderline findings go to a watch list, compliance flags state risk categories rather than predictions, and nothing is applied for you. You review everything before anything changes.

What does it cost?

Nothing. You trade an email address, we send the toolkit and its updates. It is maintained quarterly because both Amazon and Claude change under it.