Preview
What it is.
Ours, and we sell the service it runs inside. Five subagent analyses run at once against a domain, and it returns one composite score out of 100 with the action plan already sorted by priority.
Running the five in parallel is what makes a monthly score affordable per client. The design worth copying is independent checks feeding one number, because that keeps month-over-month comparison honest.
What you get.
- One composite score out of 100 across five separate analyses
- A prioritized action plan, sorted before you read it
- Five checks run at once, so the audit fits inside a working session
- A repeatable score you can compare month over month
How to set it up.
Split the audit into independent checks before you build anything. Ours runs five, and independence is what allows the parallel run.
Give each subagent one area and one output shape, so the composite score can add them up.
Score out of 100 and keep the scale identical between runs. A moving scale makes month-over-month comparison meaningless.
Sort the output by priority inside the skill, not afterwards in a spreadsheet.
Run it monthly against the same domain, because a single AI-search reading is one day of one assistant.
If you want an installable version of this job, seo-geo and ai-citations-report cover parts of it from public repos.
Pricing Plans
Not sold as a skill.
It runs inside client engagements. The published account dates from April 2026.
Use cases
The monthly client score
Each client wants an AI-search number they can watch. Run the five checks against the domain; one composite score comes back with the action plan already sorted, and the parallel run keeps the frequency affordable.
Speccing an in-house version
You're building your own audit and need the architecture. The published shape, independent checks feeding one score, is most of the spec; give each subagent one area and one output shape so the composite can add them up.
Skipping the prioritization meeting
Audits usually end with a debate about what comes first. This one arrives sorted by priority, so the first work item is the first line of the plan.
Best for
A monthly AI-search score per client
One number that moves is easier to report than five separate findings lists, and the parallel design makes the frequency affordable.
Designing your own audit skill
Five independent checks feeding one composite score is the architecture, and it is published.
Deciding what to work on first
The plan arrives sorted, which is the step most audits leave to the reader.