A ComputeDriven project live_local That rung covers this page and the gate that checks it — node launch-gate.mjs re-derives every count here from the markup and refuses on a mismatch. It covers nothing on the 25 domains listed below, and this revision has not been served from wrand.cc yet.
The question this site exists to answer

How many of these connections are real?

This is a register of 25 domains and the 56 connections recorded between them. 6 of those connections are wired in code that runs — one repository importing another, an install step calling another product's function. The rest are specified, merely stated, or nothing more than common ownership. Every row below says which of the four it is, and why. The ratio is the honest part, and it is why the tiers are shown rather than averaged into a number.

Read the inventory How it was checked
25
Domains listed
56
Recorded connections
6
Wired in code
0
Evidence rungs recorded
0
Connections executed here
Four kinds of connection

Only one of the four is evidence.

A portfolio diagram with undifferentiated arrows is a wish list. These four tiers are the whole point of the page: they separate what has been built from what has been written down, and from what has merely been said. Each connection in the inventory carries exactly one of them.

Wired in code

wired in code

One codebase imports, calls or hard-fails on another. An install step that calls Delegatic.authorize/1 and refuses on denial is this tier; a shared adjective is not.

Count: 6 of 56. Checked by: reading source in another tree, not by running it.

In the spec

in the spec

A normative document names the other product and says what crosses the boundary. Real, and worth recording — but a specification is not an implementation and this page does not let the two share a colour.

Count: 17 of 56. Checked by: reading the spec that names it.

Stated only

stated only

A page says the relationship exists and nothing else does. This is the tier that catches marketing copy, and several of these were written by us about ourselves.

Count: 6 of 56. Checked by: nothing. That is what the tier means.

Ownership

ownership

The same person owns both. It explains why two domains appear in one portfolio and it establishes nothing technical whatsoever. It is the largest tier here, and saying so is the point.

Count: 27 of 56. Checked by: the registrar.
The register

All 25 domains, and every connection between them.

This is the whole thing, as ordinary markup. No request, no hydration, no framework: curl reaches every domain name, every tagline, every connection and every reason on this page. The graph a browser draws is parsed out of exactly these elements.

Every card carries a rung chip, and every one of the 25 reads ?. Nothing in this tree records a per-domain evidence rung, so none is asserted, and the publication gate refuses any page where one has been filled in from a planning document.

wired in code in the spec stated only ownership ?

ampersandboxdesign.com?[&] Ampersand Box

Holding company. box-and-box — the governance kernel: feasible ▸ permitted ▸ best, over an un-weakenable safety floor.

group
hub
owned by
traaviis.com, wrand.cc
destination
https://ampersandboxdesign.com

Connects to (13)

computedriven.com?ComputeDriven

Systems studio. T&R — a FreeBSD 15 distro assembled from pkgbase — and RRABBIT.

group
hub
owned by
ampersandboxdesign.com, wrand.cc
destination
https://computedriven.com

Connects to (3)

  • traaviis.com wired in code 0.80 The trvs stack ships inside the T&R image — verified on a pristine build.
  • specprompt.com stated only 0.80 “The specification layer of the ComputeDriven stack.” Unpushed, and no other site says so yet.
  • fleetprompt.com stated only 0.80 “The distribution layer of the ComputeDriven stack.” Unpushed, and no other site says so yet.

specprompt.com?SpecPrompt

A spec you can checksum — content-addressed spec identity. Partly built.

group
agent
owned by
— root of the graph
destination
https://specprompt.com

Connects to (1)

  • agentelic.com wired in code 0.90 Agentelic's parser consumes a SPEC.md into a SpecPrompt.Spec; its test DSL compiles the acceptance tests.

fleetprompt.com?FleetPrompt

Planned registry for evidence-bundled agents. Install = replay. One entry today.

group
agent
owned by
— root of the graph
destination
https://fleetprompt.com

Connects to (4)

  • delegatic.com wired in code 0.90 Install step 4 calls Delegatic.authorize/1 — a policy denial hard-fails the install. Git dependency.
  • opensentience.org wired in code 0.85 Install step 5 proves the manifest is Harness-loadable via OpenSentience.Harness.start_session/1. Git dependency.
  • graphonomous.com wired in code 0.80 Install step 6 initializes a memory telespace through KILN's GraphonomousClient.
  • specprompt.com in the spec 0.70 Every listing is expected to carry a spec.

webhost.systems?WebHost.Systems

An agent control plane for the [&] Protocol. Implemented and tested in-tree; it does not yet execute an agent on any runtime.

group
infra
owned by
ampersandboxdesign.com
destination
https://webhost.systems

Connects to (2)

  • fleetprompt.com in the spec 0.70 Spec declares deploy_to_fleetprompt — a ConsolidationEvent carrying usage-based trust signals.
  • opensentience.org in the spec 0.70 Runtime follows the OS-010 PULSE specification.

bendscript.com?BendScript

Graph-first document protocol. Typed inline link facets, content-addressable ids.

group
lang
owned by
ampersandboxdesign.com
destination
https://bendscript.com

Connects to (3)

  • graphonomous.com in the spec 0.50 Facet bendscript.memory.v1 targets Graphonomous-style memory nodes.
  • specprompt.com in the spec 0.50 Facet bendscript.spec.v1 targets SpecPrompt-style capability specifications.
  • runefort.com in the spec 0.40 Facet bendscript.runefort.v1 — documents claimed by Runefort rooms.

alkeyword.com?Alkeyword

Transmutes a domain into AI-search visibility. Crawls a site, measures what AI engines say about it, turns its own claims into research briefs.

group
lang
owned by
wrand.cc
destination
https://alkeyword.com

Connects to (0)

    wrand.cc?Wrand.cc

    Personal brand. Forking. You are here.

    group
    brand
    owned by
    — root of the graph
    destination
    https://wrand.cc

    Connects to (10)

    How this was checked

    A map that cannot be read without a browser is not a map.

    Everything printed on this page is derived from the inventory above, at load, by re-reading the markup. The counts are not typed in; they are recomputed and compared, and a disagreement paints a red banner across the top of the page rather than passing quietly. launch-gate.mjs does the same pass over the emitted file and refuses to publish on a mismatch.

    Status of this surface

    Status
    live_local — the inventory on this page parses, counts and cross-checks in a browser on a developer machine, and the publication gate runs against the emitted file. Nothing here executes any of the connections it describes.
    Last verified
    2026-08-16
    Source
    node launch-gate.mjs in this repository, run against the emitted index.html; and checkDrift() in the page itself, which re-derives every printed count from the markup at load.
    Limit
    This establishes that the page is internally consistent — the counts, the JSON-LD, the tier totals and the graph all come from the same 25 articles. It does not establish that any connection was executed. The four evidence tiers were assigned by reading source in a different repository, and nothing on this page re-checks them; a connection labelled wired in code is a claim about code you would have to go and read. Nor does the HTTP 200 sweep below establish that anything is behind a domain — only that a server answered on the day it ran.
    Next rung
    live_deployed — push this revision and re-read https://wrand.cc/. The deployed copy is still the previous one, so nothing stated here has yet been checked on the domain it describes.

    What has actually been measured, and what is still zero

    Domains answering
    25 of 25 returned HTTP 200 to a plain curl on 2026-08-16. That is the only status claim on this page that has been measured. It says nothing about what is behind each domain, and it moves only in a commit that also moves that date.
    Extractable text
    Over 10,000 characters of this page survive with <script> and <style> stripped and tags removed. The previous revision of this site yielded 157. The gate refuses to publish below the floor, so the failure that produced 157 cannot recur silently.
    Evidence rungs recorded
    0. Nothing in this tree records a per-domain evidence rung, so all 25 chips read ? and the gate refuses any page where one of them has been filled in from a planning document. A missing status can still be asked about; an invented one cannot.
    Connections executed here
    0. Nothing on this page runs any of the 56. The tiers are a reading of source, and the honest way to check one is to open the file it names.

    One claim corrected in this revision

    This page described webhost.systems as a “multi-runtime AI agent deployment platform”. That claim was taken down on its own site in the same week: the Cloudflare deploy path there is a // TODO that writes simulated: true, AWS Bedrock AgentCore is a value in a dropdown with nothing behind it, and the control plane named a database the monorepo does not depend on. The card now says what that site now says — an agent control plane, implemented and tested in-tree, which does not yet execute an agent on any runtime. Naming the wrong description is what a correction is, so it stays written here rather than being quietly refreshed.

    What you can do here

    Two rungs, two different invitations.

    A page may only ask you to do what its evidence has earned. The inventory has been parsed and checked, so it asks you to read it and to re-run the check. The rung register does not exist, so it cannot ask you to run anything at all.

    live_local — parsed and cross-checked
    Use itThe register is on this page as plain markup. Every domain, every connection, every reason — readable with JavaScript switched off, and greppable with curl alone. Reproduce it locallyClone, then node launch-gate.mjs. No dependencies. It re-derives every count on this page from the markup and prints its own failures.
    spec — the rung register is not built
    ReadThe page and its gate are one file and one script. What this site claims and what it checks are the same list. ChallengeFind a connection labelled wired in code that is not. An import that does not exist is the most useful thing anyone can send. ImplementA per-domain rung source. Every chip here reads ? because nothing in the tree records one, and this page will not invent one to look finished.
    Tell us a connection is wrong

    An edge labelled wired in code that is not wired is the most useful thing you can send.

    Every tier on this page was assigned by reading source in the repository the connection names. Reading source is fallible and some of it was read quickly. If you find an import that does not exist, a spec that does not say what a row claims, or a number here that disagrees with the tree, this goes straight to a person.

    It posts to formspree.io when you press the button, carrying what you typed and nothing else. Nothing else on this page is fetched at runtime. If you would rather leave a public trace, the issue tracker is the other route and it works just as well.