Twenty one things that would otherwise be assertions
A crawler that crawls this site while you watch. A retrieval scorer that ranks this page against your question. An entity graph that reads its own markup. Each component here exists because building the claim is more convincing than writing it.
All components
Self Inspecting Entity Graph
Parses the JSON-LD out of its own document head and draws the result. The artifact and the demonstration are the same file.
Live Core Web Vitals
Measures the session you are in right now, in your browser, on your device. No vendor script and no beacon.
Raw Response vs Rendered DOM
A wipe between the bytes a crawler receives and the DOM a browser builds, with a real node and link level diff.
Attribution Ledger
Every client site currently carrying the credit link, sized by real page count and grouped by sector.
The Machine Mirror
How a language model describes an entity, diffed against the canonical claims that entity actually publishes.
Four Surface Preview
One page rendered simultaneously as a blue link, an AI Overview citation, a chat source card, and a Perplexity reference.
The Cite Block
A canonical entity record published in four serialisations at once, every one projected from the same source object.
Skill Runner
Five real audit skills running client side against any HTML you paste, including this page's own source.
Branded Report Renderer
Toggle findings and watch a client deliverable rebuild live, in the same document system used for the PDFs.
Build Receipts
The full pipeline behind a 126 page client site, from public research through sampled brand tokens to pre publish gates.
Operator Spec Sheet
Working style, operating parameters, and honest failure modes, published as a technical datasheet rather than a bio.
Career as Commit History
A professional timeline rendered as a git log, built from real repository creation dates spanning 2022 to now.
SEO ROI Against Passive Returns
The same money in an index fund instead. Return on spend, maximum drawdown, and a range of outcomes rather than one confident line.
Live Self Crawl
Reads this site's own sitemap, fetches a sample in your browser, builds the real internal link graph and runs PageRank across it.
Passage Retrieval Scorer
Chunks this page the way a retrieval pipeline would, scores each passage with real BM25, and checks whether the winner stands alone.
Crawler Access Matrix
Evaluates the live robots.txt against 27 named crawlers using the actual group and longest match precedence rules.
Declared vs Computed Topic
Real TF-IDF over the body text against a corpus from this site, compared with what the title, headings, and schema claim.
Rich Result Eligibility
Walks every typed node including nested ones and reports which enhancements it qualifies for and exactly what is blocking the rest.
Crawler Cost Meter
What a non rendering crawler downloads versus what a browser downloads to render the same page, in real measured bytes.
Near Duplicate Detector
Fetches real pages, strips the shared template, and measures what is left with SimHash fingerprints and Jaccard shingling.
Page Passport
One script tag on every page. Five analysis surfaces, run on demand, costing nothing at all until someone opens it.
Why these exist
A portfolio can claim technical fluency in a paragraph. It is harder, and considerably more convincing, to build the thing being claimed and let someone poke at it. Each component here demonstrates something that would otherwise be an assertion.
Nothing is faked
Where a component uses real data it says so, and where a live API is unavailable on a static site it says that too. The recorded runs are labelled as recorded.
No dependencies
Every component is vanilla JavaScript with inline styles. No framework, no CDN, no build step. That is a constraint worth demonstrating on an SEO site.
Isolated from the main site
Nothing here modifies the global stylesheet or any existing page. Each component is self contained, which is also why any of them can be removed without consequence.