diff --git a/README.md b/README.md
index 7f1bf8d..223975f 100644
--- a/README.md
+++ b/README.md
@@ -1,4 +1,8 @@
-# agent-vm
+# BoundaryKit
+
+> **BoundaryKit** (formerly **agent-vm**) is a public reference architecture for governing untrusted autonomous workloads with explicit runtime boundaries, promotion controls, rollback paths, and evidence-backed validation.
+>
+> The project repository and public site remain hosted at `sambegui/agent-vm` and `agent-vm.sabe.dev`.
 
 **A framework-agnostic sandbox skeleton for governed AI-agent workloads that are treated as untrusted.**
 
diff --git a/docs/architecture/00-overview.md b/docs/architecture/00-overview.md
index d6b6d3f..b79462b 100644
--- a/docs/architecture/00-overview.md
+++ b/docs/architecture/00-overview.md
@@ -6,7 +6,7 @@ Most enterprise AI usage still happens through isolated browser tabs or ad hoc a
 fragments context, hides execution state, and gives reviewers little evidence about what ran, what
 authority was available, or how recovery would work.
 
-`agent-vm` presents a governed runtime architecture for autonomous AI-agent workloads that are
+`BoundaryKit` presents a governed runtime architecture for autonomous AI-agent workloads that are
 treated as untrusted. Agents can process approved messages, attached documents, external links,
 transcripts, API responses, and structured outputs, but the platform does not trust the agent or the
 inputs it receives.
diff --git a/docs/evidence/governed-agent-workload-case-study.md b/docs/evidence/governed-agent-workload-case-study.md
index 63b049c..78d1b67 100644
--- a/docs/evidence/governed-agent-workload-case-study.md
+++ b/docs/evidence/governed-agent-workload-case-study.md
@@ -1,6 +1,6 @@
 # Case Study: Treating an Agent as an Untrusted Workload
 
-This case study is an illustrative, vendor-neutral walkthrough for reading the `agent-vm`
+This case study is an illustrative, vendor-neutral walkthrough for reading the `BoundaryKit`
 architecture. It is not a product claim, customer deployment claim, or proof that a production
 workload is ready. Its purpose is to show how the repository's trust boundaries, operational
 controls, and evidence model apply when an autonomous AI-agent workload handles untrusted input.
diff --git a/platform/tests/docs-public-contract-test b/platform/tests/docs-public-contract-test
index eea9482..1b1dabb 100644
--- a/platform/tests/docs-public-contract-test
+++ b/platform/tests/docs-public-contract-test
@@ -94,9 +94,9 @@ require_text "$ROOT/site/index.html" 'rel="canonical"'
 require_text "$ROOT/site/index.html" 'property="og:title"'
 require_text "$ROOT/site/index.html" 'name="twitter:card"'
 require_text "$ROOT/site/index.html" 'aria-controls="nav-links"'
-require_text "$ROOT/site/docs/architecture/00-overview.html" "<title>00 — Architecture overview · agent-vm specs</title>"
+require_text "$ROOT/site/docs/architecture/00-overview.html" "<title>00 — Architecture overview · BoundaryKit specs</title>"
 require_text "$ROOT/site/docs/architecture/00-overview.html" '<meta name="description"'
-require_text "$ROOT/site/docs/evidence/governed-agent-workload-case-study.html" "<title>Case Study: Treating an Agent as an Untrusted Workload · agent-vm specs</title>"
+require_text "$ROOT/site/docs/evidence/governed-agent-workload-case-study.html" "<title>Case Study: Treating an Agent as an Untrusted Workload · BoundaryKit specs</title>"
 require_text "$ROOT/docs/operations/operator-quickstart.md" "reference implementation + acceptance suite"
 require_text "$ROOT/docs/operations/operator-quickstart.md" "make provision"
 require_text "$ROOT/docs/operations/operator-quickstart.md" "AGENT_VM_SSH"
diff --git a/site/assets/trust-boundary.svg b/site/assets/trust-boundary.svg
index 3d1e26a..5cbdb94 100644
--- a/site/assets/trust-boundary.svg
+++ b/site/assets/trust-boundary.svg
@@ -1,5 +1,5 @@
 <svg width="980" height="720" viewBox="0 0 980 720" role="img" xmlns="http://www.w3.org/2000/svg">
-  <title>agent-vm trust-boundary architecture</title>
+  <title>BoundaryKit trust-boundary architecture</title>
   <desc>Grounded trust boundary flow from host operators to control plane and isolation substrate with governance gates.</desc>
   <defs>
     <linearGradient id="panel" x1="0" x2="1" y1="0" y2="1">
diff --git a/site/build_docs.py b/site/build_docs.py
index f873326..5f0e45f 100755
--- a/site/build_docs.py
+++ b/site/build_docs.py
@@ -16,7 +16,7 @@
   <meta charset="UTF-8">
   <meta name="viewport" content="width=device-width, initial-scale=1.0">
   <meta name="description" content="{description}">
-  <title>{title} · agent-vm specs</title>
+  <title>{title} · BoundaryKit specs</title>
   <style>
     :root {{
       color-scheme: dark;
@@ -307,7 +307,7 @@
     </main>
 
     <footer>
-      agent-vm security substrate · open-source verification receipt
+      BoundaryKit security substrate · open-source verification receipt
     </footer>
   </div>
 </body>
diff --git a/site/docs/architecture/00-overview.html b/site/docs/architecture/00-overview.html
index f341d26..ea62747 100644
--- a/site/docs/architecture/00-overview.html
+++ b/site/docs/architecture/00-overview.html
@@ -4,7 +4,7 @@
   <meta charset="UTF-8">
   <meta name="viewport" content="width=device-width, initial-scale=1.0">
   <meta name="description" content="Most enterprise AI usage still happens through isolated browser tabs or ad hoc automation. That model fragments context, hides execution state, and gives reviewers little...">
-  <title>00 — Architecture overview · agent-vm specs</title>
+  <title>00 — Architecture overview · BoundaryKit specs</title>
   <style>
     :root {
       color-scheme: dark;
@@ -296,7 +296,7 @@ <h2 id="shifting-from-chatbot-tabs-to-governed-persistent-agent-workloads">Shift
 <p>Most enterprise AI usage still happens through isolated browser tabs or ad hoc automation. That model
 fragments context, hides execution state, and gives reviewers little evidence about what ran, what
 authority was available, or how recovery would work.</p>
-<p><code>agent-vm</code> presents a governed runtime architecture for autonomous AI-agent workloads that are
+<p><code>BoundaryKit</code> presents a governed runtime architecture for autonomous AI-agent workloads that are
 treated as untrusted. Agents can process approved messages, attached documents, external links,
 transcripts, API responses, and structured outputs, but the platform does not trust the agent or the
 inputs it receives.</p>
@@ -367,7 +367,7 @@ <h2 id="how-to-read-this-directory">How to read this directory</h2>
     </main>
 
     <footer>
-      agent-vm security substrate · open-source verification receipt
+      BoundaryKit security substrate · open-source verification receipt
     </footer>
   </div>
 </body>
diff --git a/site/docs/architecture/01-isolation-substrate.html b/site/docs/architecture/01-isolation-substrate.html
index ebd1f86..78bd01f 100644
--- a/site/docs/architecture/01-isolation-substrate.html
+++ b/site/docs/architecture/01-isolation-substrate.html
@@ -4,7 +4,7 @@
   <meta charset="UTF-8">
   <meta name="viewport" content="width=device-width, initial-scale=1.0">
   <meta name="description" content="The substrate is a golden VM (agent-platform) defined as code on a nested-virtualization host. Agents never run on the host directly; the VM is the first boundary, and inside...">
-  <title>01 — Isolation substrate · agent-vm specs</title>
+  <title>01 — Isolation substrate · BoundaryKit specs</title>
   <style>
     :root {
       color-scheme: dark;
@@ -391,7 +391,7 @@ <h2 id="provisioned-as-code">Provisioned as code</h2>
     </main>
 
     <footer>
-      agent-vm security substrate · open-source verification receipt
+      BoundaryKit security substrate · open-source verification receipt
     </footer>
   </div>
 </body>
diff --git a/site/docs/architecture/02-promotion-control-plane.html b/site/docs/architecture/02-promotion-control-plane.html
index f8cca47..eb58bef 100644
--- a/site/docs/architecture/02-promotion-control-plane.html
+++ b/site/docs/architecture/02-promotion-control-plane.html
@@ -4,7 +4,7 @@
   <meta charset="UTF-8">
   <meta name="viewport" content="width=device-width, initial-scale=1.0">
   <meta name="description" content="The control plane is a host-side set of small, idempotent scripts plus versioned Markdown &quot;truth&quot; files. It deploys agent runtimes over ssh against immutable releases — live...">
-  <title>02 — Promotion control plane · agent-vm specs</title>
+  <title>02 — Promotion control plane · BoundaryKit specs</title>
   <style>
     :root {
       color-scheme: dark;
@@ -382,7 +382,7 @@ <h2 id="library">Library</h2>
     </main>
 
     <footer>
-      agent-vm security substrate · open-source verification receipt
+      BoundaryKit security substrate · open-source verification receipt
     </footer>
   </div>
 </body>
diff --git a/site/docs/architecture/03-gateway-runtime-layout.html b/site/docs/architecture/03-gateway-runtime-layout.html
index dbdf7a4..8c413de 100644
--- a/site/docs/architecture/03-gateway-runtime-layout.html
+++ b/site/docs/architecture/03-gateway-runtime-layout.html
@@ -4,7 +4,7 @@
   <meta charset="UTF-8">
   <meta name="viewport" content="width=device-width, initial-scale=1.0">
   <meta name="description" content="A single agent runtime often serves several profiles at once (a primary channel gateway, an HTTP API surface, an MCP/SSE endpoint, …). The failure this layer prevents: many...">
-  <title>03 — Gateway runtime layout · agent-vm specs</title>
+  <title>03 — Gateway runtime layout · BoundaryKit specs</title>
   <style>
     :root {
       color-scheme: dark;
@@ -408,7 +408,7 @@ <h2 id="alignment-proof">Alignment proof</h2>
     </main>
 
     <footer>
-      agent-vm security substrate · open-source verification receipt
+      BoundaryKit security substrate · open-source verification receipt
     </footer>
   </div>
 </body>
diff --git a/site/docs/architecture/04-production-governance.html b/site/docs/architecture/04-production-governance.html
index 9301eb4..62c296a 100644
--- a/site/docs/architecture/04-production-governance.html
+++ b/site/docs/architecture/04-production-governance.html
@@ -4,7 +4,7 @@
   <meta charset="UTF-8">
   <meta name="viewport" content="width=device-width, initial-scale=1.0">
   <meta name="description" content="The substrate makes workloads isolatable; governance decides what a given workload is allowed to be, and gates it from lab to production with controls proportional to its blast...">
-  <title>04 — Production governance · agent-vm specs</title>
+  <title>04 — Production governance · BoundaryKit specs</title>
   <style>
     :root {
       color-scheme: dark;
@@ -491,7 +491,7 @@ <h2 id="production-ready-means-evidence-not-unit-creation">Production-ready mean
     </main>
 
     <footer>
-      agent-vm security substrate · open-source verification receipt
+      BoundaryKit security substrate · open-source verification receipt
     </footer>
   </div>
 </body>
diff --git a/site/docs/architecture/05-secure-gated-agent-preview-access.html b/site/docs/architecture/05-secure-gated-agent-preview-access.html
index aadfcbd..fe38005 100644
--- a/site/docs/architecture/05-secure-gated-agent-preview-access.html
+++ b/site/docs/architecture/05-secure-gated-agent-preview-access.html
@@ -4,7 +4,7 @@
   <meta charset="UTF-8">
   <meta name="viewport" content="width=device-width, initial-scale=1.0">
   <meta name="description" content="Agent previews are useful because reviewers need to see the actual running surface, not only a screenshot or a local build log. They are risky because an agent runtime or...">
-  <title>05 — Secure gated agent preview access · agent-vm specs</title>
+  <title>05 — Secure gated agent preview access · BoundaryKit specs</title>
   <style>
     :root {
       color-scheme: dark;
@@ -674,7 +674,7 @@ <h2 id="non-goals">Non-goals</h2>
     </main>
 
     <footer>
-      agent-vm security substrate · open-source verification receipt
+      BoundaryKit security substrate · open-source verification receipt
     </footer>
   </div>
 </body>
diff --git a/site/docs/evidence/governed-agent-workload-case-study.html b/site/docs/evidence/governed-agent-workload-case-study.html
index f1e617a..c94b15d 100644
--- a/site/docs/evidence/governed-agent-workload-case-study.html
+++ b/site/docs/evidence/governed-agent-workload-case-study.html
@@ -3,8 +3,8 @@
 <head>
   <meta charset="UTF-8">
   <meta name="viewport" content="width=device-width, initial-scale=1.0">
-  <meta name="description" content="This case study is an illustrative, vendor-neutral walkthrough for reading the agent-vm architecture. It is not a product claim, customer deployment claim, or proof that a...">
-  <title>Case Study: Treating an Agent as an Untrusted Workload · agent-vm specs</title>
+  <meta name="description" content="This case study is an illustrative, vendor-neutral walkthrough for reading the BoundaryKit architecture. It is not a product claim, customer deployment claim, or proof that a...">
+  <title>Case Study: Treating an Agent as an Untrusted Workload · BoundaryKit specs</title>
   <style>
     :root {
       color-scheme: dark;
@@ -292,7 +292,7 @@
     </header>
     <main>
       <h1 id="case-study-treating-an-agent-as-an-untrusted-workload">Case Study: Treating an Agent as an Untrusted Workload</h1>
-<p>This case study is an illustrative, vendor-neutral walkthrough for reading the <code>agent-vm</code>
+<p>This case study is an illustrative, vendor-neutral walkthrough for reading the <code>BoundaryKit</code>
 architecture. It is not a product claim, customer deployment claim, or proof that a production
 workload is ready. Its purpose is to show how the repository's trust boundaries, operational
 controls, and evidence model apply when an autonomous AI-agent workload handles untrusted input.</p>
@@ -448,7 +448,7 @@ <h2 id="reviewer-takeaway">Reviewer takeaway</h2>
     </main>
 
     <footer>
-      agent-vm security substrate · open-source verification receipt
+      BoundaryKit security substrate · open-source verification receipt
     </footer>
   </div>
 </body>
diff --git a/site/docs/evidence/substrate-validation-receipt.html b/site/docs/evidence/substrate-validation-receipt.html
index db9fccb..8742d72 100644
--- a/site/docs/evidence/substrate-validation-receipt.html
+++ b/site/docs/evidence/substrate-validation-receipt.html
@@ -4,7 +4,7 @@
   <meta charset="UTF-8">
   <meta name="viewport" content="width=device-width, initial-scale=1.0">
   <meta name="description" content="This receipt summarizes the public, non-secret evidence recorded in platform/state/substrate-validation.md. It is a sanitized public evidence packet, not a live infrastructure log.">
-  <title>Substrate validation receipt · agent-vm specs</title>
+  <title>Substrate validation receipt · BoundaryKit specs</title>
   <style>
     :root {
       color-scheme: dark;
@@ -405,7 +405,7 @@ <h2 id="follow-up-evidence-needed-for-production-readiness">Follow-up evidence n
     </main>
 
     <footer>
-      agent-vm security substrate · open-source verification receipt
+      BoundaryKit security substrate · open-source verification receipt
     </footer>
   </div>
 </body>
diff --git a/site/docs/llms-full.txt b/site/docs/llms-full.txt
index 26e1984..abca8c3 100644
--- a/site/docs/llms-full.txt
+++ b/site/docs/llms-full.txt
@@ -1,7 +1,7 @@
-# agent-vm full LLM guide
+# BoundaryKit full LLM guide
 
 Purpose:
-agent-vm documents a public, agent-agnostic reference architecture and validation suite for running AI-agent workloads in governed, isolated runtime layers. It separates public design claims, static checks, host-dependent substrate checks, and production-readiness evidence.
+BoundaryKit (formerly agent-vm) is a public reference architecture for governing untrusted autonomous workloads with explicit runtime boundaries, promotion controls, rollback paths, and evidence-backed validation. It documents a public, agent-agnostic reference architecture and validation suite for running AI-agent workloads in governed, isolated runtime layers. It separates public design claims, static checks, host-dependent substrate checks, and production-readiness evidence.
 
 Primary pages:
 - / : Public landing page with the trust-boundary map, validated scope, and evidence links.
diff --git a/site/docs/llms.txt b/site/docs/llms.txt
index 0208446..f5b6905 100644
--- a/site/docs/llms.txt
+++ b/site/docs/llms.txt
@@ -1,6 +1,6 @@
-# agent-vm LLM guide
+# BoundaryKit LLM guide
 
-agent-vm is a public reference implementation and validation suite for governing untrusted autonomous AI-agent workloads inside isolated runtime layers. Treat it as a reference architecture and reproducible skeleton, not as a claim that every production operation is turnkey.
+BoundaryKit (formerly agent-vm) is a public reference architecture for governing untrusted autonomous workloads with explicit runtime boundaries, promotion controls, rollback paths, and evidence-backed validation. It is a reference implementation and validation suite for governing untrusted autonomous AI-agent workloads inside isolated runtime layers — treat it as a reproducible skeleton, not as a claim that every production operation is turnkey.
 
 Start here:
 - Home: /
diff --git a/site/docs/security-methodology.html b/site/docs/security-methodology.html
index a11b4b7..5319498 100644
--- a/site/docs/security-methodology.html
+++ b/site/docs/security-methodology.html
@@ -4,7 +4,7 @@
   <meta charset="UTF-8">
   <meta name="viewport" content="width=device-width, initial-scale=1.0">
   <meta name="description" content="This document describes the security methodology used to design and review the reference architecture. It applies to the walking-skeleton code in platform/ and control-plane/.">
-  <title>Security methodology · agent-vm specs</title>
+  <title>Security methodology · BoundaryKit specs</title>
   <style>
     :root {
       color-scheme: dark;
@@ -423,7 +423,7 @@ <h3 id="post-session">Post-session</h3>
     </main>
 
     <footer>
-      agent-vm security substrate · open-source verification receipt
+      BoundaryKit security substrate · open-source verification receipt
     </footer>
   </div>
 </body>
diff --git a/site/docs/threat-model.html b/site/docs/threat-model.html
index 99186fc..5737069 100644
--- a/site/docs/threat-model.html
+++ b/site/docs/threat-model.html
@@ -4,7 +4,7 @@
   <meta charset="UTF-8">
   <meta name="viewport" content="width=device-width, initial-scale=1.0">
   <meta name="description" content="This reference architecture treats AI agents as untrusted, tool-wielding workloads. The platform assumes an agent can be misdirected by untrusted input or compromised through...">
-  <title>Threat model · agent-vm specs</title>
+  <title>Threat model · BoundaryKit specs</title>
   <style>
     :root {
       color-scheme: dark;
@@ -454,7 +454,7 @@ <h2 id="security-invariants">Security invariants</h2>
     </main>
 
     <footer>
-      agent-vm security substrate · open-source verification receipt
+      BoundaryKit security substrate · open-source verification receipt
     </footer>
   </div>
 </body>
diff --git a/site/docs/verification.html b/site/docs/verification.html
index 3175e21..9830670 100644
--- a/site/docs/verification.html
+++ b/site/docs/verification.html
@@ -4,7 +4,7 @@
   <meta charset="UTF-8">
   <meta name="viewport" content="width=device-width, initial-scale=1.0">
   <meta name="description" content="This repository separates static validation, host-dependent substrate validation, and production-readiness evidence. Do not collapse those into one claim.">
-  <title>Verification model · agent-vm specs</title>
+  <title>Verification model · BoundaryKit specs</title>
   <style>
     :root {
       color-scheme: dark;
@@ -386,7 +386,7 @@ <h2 id="claim-discipline">Claim discipline</h2>
     </main>
 
     <footer>
-      agent-vm security substrate · open-source verification receipt
+      BoundaryKit security substrate · open-source verification receipt
     </footer>
   </div>
 </body>
diff --git a/site/index.html b/site/index.html
index 2268a2d..3bf8d18 100644
--- a/site/index.html
+++ b/site/index.html
@@ -5,17 +5,17 @@
     <meta name="viewport" content="width=device-width, initial-scale=1" />
     <meta
       name="description"
-      content="agent-vm is a public reference architecture for governing untrusted autonomous AI-agent workloads with explicit trust boundaries, rollback, audit, and validation evidence."
+      content="BoundaryKit is a public reference architecture for governing untrusted autonomous workloads with explicit runtime boundaries, promotion controls, rollback paths, and evidence-backed validation."
     />
     <link rel="canonical" href="https://agent-vm.sabe.dev/" />
     <meta property="og:type" content="website" />
-    <meta property="og:title" content="agent-vm — governed agent workload architecture" />
-    <meta property="og:description" content="A public reference architecture for governing untrusted autonomous AI-agent workloads with explicit trust boundaries, rollback, audit, and validation evidence." />
+    <meta property="og:title" content="BoundaryKit — governed runtime boundaries for autonomous agent workloads" />
+    <meta property="og:description" content="BoundaryKit is a public reference architecture for governing untrusted autonomous workloads with explicit runtime boundaries, promotion controls, rollback paths, and evidence-backed validation." />
     <meta property="og:url" content="https://agent-vm.sabe.dev/" />
     <meta name="twitter:card" content="summary" />
-    <meta name="twitter:title" content="agent-vm — governed agent workload architecture" />
-    <meta name="twitter:description" content="A public reference architecture for governing untrusted autonomous AI-agent workloads with explicit trust boundaries, rollback, audit, and validation evidence." />
-    <title>agent-vm — governed agent workload architecture</title>
+    <meta name="twitter:title" content="BoundaryKit — governed runtime boundaries for autonomous agent workloads" />
+    <meta name="twitter:description" content="BoundaryKit is a public reference architecture for governing untrusted autonomous workloads with explicit runtime boundaries, promotion controls, rollback paths, and evidence-backed validation." />
+    <title>BoundaryKit — governed runtime boundaries for autonomous agent workloads</title>
     <link rel="stylesheet" href="styles.css" />
     <link rel="icon" href="favicon.ico" sizes="any" />
     <link rel="icon" href="favicon-32x32.png" type="image/png" />
@@ -26,17 +26,17 @@
     <header class="site-header" aria-label="Primary navigation">
       <div class="header-main-row">
         <div class="brand">
-          <a class="brand-mark" href="https://github.com/sambegui/agent-vm" target="_blank" rel="noopener noreferrer" aria-label="agent-vm repository">
-            <span class="brand-fallback">av</span>
+          <a class="brand-mark" href="https://github.com/sambegui/agent-vm" target="_blank" rel="noopener noreferrer" aria-label="BoundaryKit repository">
+            <span class="brand-fallback">bk</span>
             <img
               src="assets/agent-vm-avatar.jpg"
               alt=""
               decoding="async"
             />
           </a>
-          <a class="brand-copy" href="#top" aria-label="agent-vm home">
-            <span>agent-vm</span>
-            <small>governed agent workload architecture</small>
+          <a class="brand-copy" href="#top" aria-label="BoundaryKit home">
+            <span>BoundaryKit</span>
+            <small>Governed runtime boundaries for autonomous agent workloads</small>
           </a>
         </div>
         <button class="menu-toggle" id="menu-toggle" aria-label="Toggle navigation" aria-expanded="false" aria-controls="nav-links">
@@ -64,7 +64,7 @@
           <p class="eyebrow">governed agent workloads · trust-boundary evidence</p>
           <h1 id="hero-title">Govern untrusted AI-agent workloads with explicit boundaries and evidence.</h1>
           <p class="hero-lede">
-            <strong>agent-vm</strong> is a public reference architecture for treating autonomous AI agents as probabilistic workloads that process untrusted input. It demonstrates how promotion controls, isolation, default-deny access, audit trails, rollback, and reproducible validation fit together without claiming a turnkey security product.
+            <strong>BoundaryKit</strong> (formerly agent-vm) is a public reference architecture for treating autonomous AI agents as probabilistic workloads that process untrusted input. It demonstrates how promotion controls, isolation, default-deny access, audit trails, rollback, and reproducible validation fit together without claiming a turnkey security product.
           </p>
           <div class="hero-actions">
             <a class="button primary" href="#architecture">See the boundary</a>
diff --git a/site/llms.txt b/site/llms.txt
index 0208446..f5b6905 100644
--- a/site/llms.txt
+++ b/site/llms.txt
@@ -1,6 +1,6 @@
-# agent-vm LLM guide
+# BoundaryKit LLM guide
 
-agent-vm is a public reference implementation and validation suite for governing untrusted autonomous AI-agent workloads inside isolated runtime layers. Treat it as a reference architecture and reproducible skeleton, not as a claim that every production operation is turnkey.
+BoundaryKit (formerly agent-vm) is a public reference architecture for governing untrusted autonomous workloads with explicit runtime boundaries, promotion controls, rollback paths, and evidence-backed validation. It is a reference implementation and validation suite for governing untrusted autonomous AI-agent workloads inside isolated runtime layers — treat it as a reproducible skeleton, not as a claim that every production operation is turnkey.
 
 Start here:
 - Home: /