When we published the AEO Standard two weeks ago, the announcement carried an honest asterisk: the rubric was public, but the scanner that enforces it in our CI pipelines was not. A standard you can read but cannot run is a document; a standard with a reference implementation is a tool other teams can adopt, test, and argue with. Today the asterisk is gone.
aeo-scan is open source — MIT-licensed, dependency-free, and public on GitHub — and it did not travel alone.
In the same push, our technical whitepapers gained a verbatim mirror repository, the standard's reference repo now links its implementation, and isimplifyme.com's llms.txt was accepted into the largest community directory of llms.txt implementations. What follows is what each piece is, why we opened it, and how to put the scanner to work on your own site this afternoon.
What is aeo-scan?
aeo-scan is iSimplifyMe's open-source scanner for Answer Engine Optimization, published on GitHub under an MIT license. It audits pages against the mechanical checks of the AEO Standard in two modes — source-parse for Next.js App Router projects and fetch mode for any rendered HTML — with JSON-LD schema validation and CI exit codes, using zero runtime dependencies.
What Actually Shipped
Four things, each small on its own, deliberately wired together:
- The scanner, open. The aeo-scan repository is public under MIT — the same package that gates our own deploys. Source-parse mode walks a Next.js App Router tree and checks every page for the structural elements the standard requires; fetch mode audits rendered HTML for title and description length bands, canonical tags, heading structure, atomic answer blocks, and JSON-LD schema with per-type required-field validation. It ships with its test suite and no runtime dependencies at all.
- The whitepapers, mirrored. All seven of our published papers — from the healthcare Bedrock reference architecture to the Trust Ladder — now live in a public mirror repository under CC BY 4.0, each carrying a canonical link back to its living edition. The papers revise on the domain first; the mirror follows.
- The directory listing. The llms-txt-hub — the largest community directory of sites implementing the llms.txt convention — accepted our submission, so machines discovering AI-readable sites through that index now find ours in it.
- The wiring. The standard links the scanner, the scanner links the standard, the mirror links the canon, and the npm preview links all three. None of these artifacts is an orphan.
Why open-source an AEO scanner?
Because a published standard without a runnable implementation is unverifiable. Open-sourcing aeo-scan lets any team check what the AEO Standard's mechanical checks actually measure, reproduce a score, and dispute a threshold with evidence. It applies the standard's own first principle to the tooling: claims you cannot inspect deserve no authority.
There is also a less lofty reason, and the standard obliges us to state it plainly: distribution. The heaviest section of the rubric scores original, demonstrable expertise — and engineers evaluate tools where tools live, which is a repository they can read, not a landing page. Publishing the scanner where its audience already works is the same discipline the standard teaches for content: put the substance where the reader — human or machine — actually looks.
What The Scanner Does — And Deliberately Does Not
The AEO Standard marks every check mechanical or judgment, and aeo-scan scores only the mechanical half. Length bands, canonical presence, heading hierarchy, atomic-answer counts, FAQ schema depth, breadcrumb completeness — software can verify all of it, and now anyone can verify the software.
The judgment half — all twenty-five points of Substance & Originality — stays with humans and LLM review passes, because whether a page says anything worth citing cannot be measured by regex, and a tool that pretended otherwise would reward exactly the commodity content the standard exists to filter out.
One detail from the unification is worth engineers' attention. While consolidating our per-site scanner scripts into this package, we found that the legacy versions were blind to JSON-LD's @graph container — every schema type inside {"@graph":[...]} was reported missing, including on our own homepage. The open package parses all three common JSON-LD shapes, with a regression test to keep it that way.
If your own auditing scripts predate @graph awareness, they are quietly underreporting your structured data; it is the single most common false negative we have seen.
Can aeo-scan run in CI?
Yes — CI gating is what aeo-scan was built for. It exits nonzero when a route fails its tier's required checks, classifies routes into strict, light, and policy tiers via a JSON config file, and emits machine-readable output with a JSON flag. iSimplifyMe's own site deploys run it as a blocking gate before every production push.
Per-site behavior lives in a single aeo-scan.config.json — route tiers, thresholds, component patterns, dynamic-segment substitutions — so the core stays small while site conventions stay yours. No configuration at all gets you sensible generic defaults in fetch mode.
What is the difference between the npm preview and the full CLI?
The aeo-scan package on npm is a single-URL preview: npx aeo-scan plus a URL checks core signals — title and description bands, canonical, H1 count, OG tags, schema types, FAQ depth — in seconds with no install. The full CLI in the GitHub repository adds source-parse mode, sitemap crawls, tiered CI gating, and Markdown report artifacts.
The preview stays deliberately boring: one URL in, one checklist out, exit code honest. When the full CLI lands on npm it will keep that single-URL contract intact — a promise we can make publicly now that the source sits where anyone can hold us to it.
Why A Score Is Still A Floor, Not A Forecast
Keep in mind what none of this changes: answer engines are not deterministic, and no structural score — from our scanner or anyone else's — guarantees a citation. A page that clears every mechanical check has earned extractability, not selection. That line is written into the standard, printed by the scanner itself, and repeated here because the alternative is a market of tools selling certainty that does not exist.
If a vendor promises citations from a score, you now have an open reference for what a score can honestly claim.
Where Everything Lives Now
The estate, in one place. The scanner: github.com/iSimplifyMe/aeo-scan (MIT). The standard: its canonical home in the Lab and the reference repository (CC BY 4.0), plus the white-paper edition. The papers: canonical editions on the domain and the GitHub mirror. The machine-readable index of all of it: isimplifyme.com/llms.txt. And for the quickest possible taste, the free web scanner runs the same checks in your browser against any URL, right now.
Frequently Asked Questions
Can I use aeo-scan commercially?
Yes. The scanner is MIT-licensed — use it, modify it, embed it in commercial products and client work without asking. The AEO Standard it scores against is separately licensed CC BY 4.0, which requires attribution to iSimplifyMe when you use or adapt the rubric itself.
Can I report bugs or contribute?
Issues are open on the repository and reviewed on a monthly cadence. The most useful reports are false positives or false negatives on a specific check, with the page HTML or a reduction attached. The core stays small and dependency-free by policy — site-specific customization belongs in the config file, not the core.
Doesn't giving away the scanner undercut your paid audit work?
No — it draws the line honestly. The scanner automates the mechanical half of the standard, which was never the scarce part. The paid work applies the judgment half: substance scoring, fan-out coverage across a cluster, and the infrastructure that enforces all of it continuously. An open mechanical scanner makes that boundary inspectable instead of mysterious.
Will the npm package become the full CLI?
That is the plan. The single-URL preview behavior — npx aeo-scan plus a URL — will keep working unchanged when the full CLI ships to npm; the added modes arrive behind flags rather than by breaking the existing contract. Until then, the full CLI is available from the repository directly.
Can I republish or translate the mirrored whitepapers?
Yes, with attribution. The mirror repository is licensed CC BY 4.0: share, redistribute, and adapt the papers with credit to iSimplifyMe. The canonical editions live at isimplifyme.com/whitepapers and revise there first — when a copy and the canon disagree, trust the canon.
Run It On Something Real
The fastest way to understand any scanner is to watch it grade a page you care about. Point the free web scanner at your site, or clone the repository and wire it into CI. And if what the mechanical checks reveal looks like an infrastructure problem rather than a copyediting one — that is the work we do.