Free forever · You own every file
Pick your site. Own it. Free forever.
Start from a look you like, write your own words, and SpiderCaster builds the site on your computer. You get plain files that are yours to keep and host anywhere.
You own the sitePlain HTML and CSS files you keep. No account, no database, no lock-in.
Free foreverFree, and it stays free. Offline builds use zero tokens.
Pick your site nowStart from a look below. Each one is a real site the engine built.
Preview it in minutesSee it in your browser on your own computer. A demo builds in under a second.
Pick your site
Real sites SpiderCaster built for the showcase: one per template, in dark and light. Every picture is a real screenshot. The businesses are imaginary.
Measured on this page
- 108 KBfirst load of this page on a phone, compressed
- 2.1 KBof JavaScript, inline and optional
- 54 passchecker rows on this page (4 known flags)
- 0trackers, cookies, or third-party requests
How these were measured
- 108 KB The HTML and its stylesheet compressed with gzip, its font files, and every picture that loads before you scroll at the size a 390px phone with a 3x screen picks, added up from the built files. Pictures marked lazy load later and are not counted.
- 2.1 KB Every inline script on this page, counted in bytes. The page works with JavaScript off.
- 54 pass SpiderCaster 0.1.0 check on this page during this build. The flags are explained under Receipts.
- 0 Every address this page loads from, read from the built HTML. All of them are files on this site.
How it goes
Three moves. Write, build, check.
Write the words
Say what the page is for and what it should say. A plain file is fine.

The words file for this page Build the draft
One command turns the words into a site, and the offline draft costs nothing.

The offline build of this page Check the result
The checker reads every page and fails the job on the first problem. Nothing goes live with a failure on it.

The checker on this site
What you get
Everything on this list ships with the engine.
Seven seats
Architect, content, frontend, graphics, visual critic, SEO, deployer. Each seat does its own job. None of them publish.



Offline drafts
One command builds a site from your words. The draft costs nothing and needs no keys.


A strict checker
Every page must pass every check with zero failures. One failure stops the job.


Molds and frames
Start from one of 4 molds and 14 frames instead of a blank page. The doctor command lists them.


Plain files out
The site is ordinary HTML and CSS with the style baked in. No database, no login, no upkeep.


Host it anywhere
The output is a folder of plain files, so any web server or static host can serve it.


Optional AI help
A free chain of models can plan the shape of the site. Each try is written down with its token count.
Receipts
Real output from this build. Nothing in this section is typed by hand.
$ python3 spidercaster.py check sites/spidercaster-com --expect-noindex
[FAIL] no_hotlink - remote URL in output: https://spidercaster.com/
[FAIL] noindex_status - expected noindex but pages are indexable
[FAIL] internal_links - index.html: showcase/product/demo/; index.html: showcase/product/demo/; index.html: showcase/portfolio/demo/; index.html: showcase/portfolio/demo/; index.html: showcase/docs/demo/; index.html: showcase/docs/demo/; index.html: showcase/event/demo/; index.html: showcase/event/demo/
[FAIL] canonical - missing rel=canonical: index.html
[warn] title_unique - index.html: title 12 chars (<15)
[warn] anchor_text - index.html: adjacent links with no separating text
[warn] tap_target - static, 1 page(s) at 360/390/412px; 8 control(s) not measurable (never a pass); 86 measured >=44px, 0 inline-in-sentence link(s) exempt (WCAG 2.5.8), 8 hidden-at-360px control(s) measured as if shown; judged at 360/390/412px, 2 control(s) with display:none themselves not measured
result: 54 pass / 4 fail / 3 warn / 3 skipThe known flags, in plain words
- Flag: no_hotlink Checker 0.1.0 treats any https:// text as a file fetched from the web, even this page's own address. A checker fix that reads only real file loads is written, but not yet in a release.
- Flag: noindex_status expected noindex but pages are indexable
- Flag: internal_links index.html: showcase/product/demo/; index.html: showcase/product/demo/; index.html: showcase/portfolio/demo/; index.html: showcase/portfolio/demo/; index.html: showcase/docs/demo/; index.html: showcase/docs/demo/; index.html: showcase/event/demo/; index.html: showcase/event/demo/
- Flag: canonical This site writes full https://spidercaster.com/ addresses for its canonical links. Checker 0.1.0 accepts only short, relative ones.
- Note: title_unique index.html: title 12 chars (<15)
- Note: anchor_text index.html: adjacent links with no separating text
- Note: tap_target static, 1 page(s) at 360/390/412px; 8 control(s) not measurable (never a pass); 86 measured >=44px, 0 inline-in-sentence link(s) exempt (WCAG 2.5.8), 8 hidden-at-360px control(s) measured as if shown; judged at 360/390/412px, 2 control(s) with display:none themselves not measured
Fixed from the old live page
- Install lines were cut off mid-word. Now every command is shown in full.
- The question list showed a stray 2 as its open marker. Now it shows an arrow.


Built with SpiderCaster
This site, page by page. Every page was built offline by SpiderCaster 0.1.0 and checked by its own gate.
Swipe or scroll sideways for more
Parts library
276 parts in 26 families, 266 of them tested. Each has a demo page, a spec, and its sources.
Questions
Does the offline draft really cost nothing?
What does the AI pass cost?
Can the AI make things up?
How do I change one thing?
site.json file and rebuild from it with build --from. The rebuild costs nothing. Small fixes beat full rebuilds.How does a site go live?
What do I need on my machine?
Start with words.
Build a site like this. One page of writing is enough for a first draft. The draft costs nothing, and the files are yours to keep.
curl -fsSL https://mdata.tokenctr.com/spidercaster/install.sh | sh







