SEOmator GEO Audit Skill

A GEO (Generative Engine Optimization) audit toolkit for Claude Code: a deterministic Python CLI, plus thin skills that narrate what it computes.

plugin   /plugin marketplace add seo-skills/geo-audit-skill  ->  /plugin install geo
CLI      uv tool install seomator-geo-audit                  ->  geo audit <url>

The one claim: every number in a report is reproducible from recorded evidence. The CLI computes; the model explains and prioritizes. No score is ever produced by an LLM doing arithmetic in prose.

Status: 0.6.0. Eleven commands, nine skills, six scoring categories, an envelope schema the build enforces, and audits that keep the pages they read. What stands between this and 1.0 is the practitioner eval — a human gate, by design. See the roadmap.

From SEOmator. This is SEOmator's free, MIT-licensed GEO audit. For more features, agentic ones included, sign up at seomator.com.

Install

uv tool install seomator-geo-audit      # or: pipx install seomator-geo-audit

Python 3.11 or newer. No browser required.

Quickstart

geo doctor
geo score https://example.com/pricing
geo audit https://example.com

geo score is the first success: one page, no crawl, no browser, no Claude Code, under thirty seconds. It prints a citability score out of 100 with its tier label, the signals behind it, and the fixes ranked by what they recover.

geo audit crawls the site and scores every category over it. It respects robots.txt for the links it discovers, holds to one request per second across the whole crawl, and records the result so you can reproduce it later:

geo audit https://example.com --rescore <run_id>   # recomputes, no network
geo report https://example.com --pdf               # a document you can send
geo report https://example.com --site-kind docs   # ordered for a documentation site
geo compare https://example.com                    # what changed since last time

What a report looks like: client copy and operator copy, generated from the synthetic fixture site.

Output is JSON whenever stdout is not a terminal, so the second command needs no flag in a pipeline. Progress goes to stderr, always.

Inside Claude Code

/plugin marketplace add seo-skills/geo-audit-skill
/plugin install geo
/geo:audit https://example.com

Six skills ship with the plugin: /geo:audit, /geo:citability, /geo:technical, /geo:schema, /geo:llmstxt and /geo:brand. They call the CLI and read its JSON. They never guess a number, and they never see raw page text.

What it measures today

Six categories, weighted to 100. Every signal is classified, and the class decides what it is allowed to do to a number.

Full definitions and thresholds: docs/concepts/signals.md.

Three rules that shape every number

A signal that could not be measured is never scored as a failure. It is null, completeness names it, and the score is taken over the signals that were computed. The same rule applies one level up: geo audit --only schema scores out of schema, not out of schema plus five zeroes. A category whose input you did not supply — brand, without a name — is out of scope rather than missing.

Model judgement never becomes a number. Two content questions are marked advisory: they carry a rubric and no value, a model answers them, and the composite filters on signal class so no code path turns an answer into a score. They appear in their own labelled section of the report.

A fact is scored once. No signal id appears in two categories, and a test asserts it. Crawler access is technical; answer-shaped schema types are schema; preview cards are platform.

Reading the output

GEO citability score 62/100 (Fair) for example.com — 1 page, evidence CURRENT.

• Number and label always travel together. Colour is a third channel on top of both, never the only one carrying the verdict.
• evidence carries a SHA-256 over the extracted content blocks plus the normalizer version, so a later run can tell you whether the page it scored is still the page you are looking at.
• Scores compare only with themselves. A score is comparable with another score from this tool at the same scoring_version and data_version. Numbers from other GEO tools measure different things and are not interchangeable.

Safety

The core loop feeds untrusted web content toward an agent with tool access, so:

• CLI output contains derived signals and length-capped, delimiter-escaped excerpts. It never contains page text.
• Redirects and links to private, loopback, link-local and metadata addresses are refused. A private start URL needs --allow-private, because auditing localhost:3000 is a legitimate thing to want.
• The address we actually connected to is validated, not only the name we resolved.

Crawls identify themselves as SeomatorGeoAudit/<version> with a link back to this repository, hold to one request per second across the whole crawl, and respect robots.txt for every link they discover.

Details and the threat model: SECURITY.md.

Exit codes

A 403 or a 5xx from the page is a finding, not a failure. Exiting non-zero there would abort exactly the sites that most need a report.

Roadmap

The gate on 1.0 is not a feature. It is the eval: two rounds where an outside practitioner would send at least four of five reports unedited.

Docs

• Quickstart · Practitioner eval
• Command reference (generated from the parser)
• Signals · Evidence · Scoring methodology · Score divergence
• Envelope JSON Schema — every command's output validates against it in CI
• Troubleshooting — every exit code and GEO_E_* code mapped to a fix
• Contributing

License

MIT, © 2026 SEOmator. See LICENSE.