diff --git a/assets/counterproof-hero.svg b/assets/counterproof-hero.svg
index 6cad989..ee48580 100644
--- a/assets/counterproof-hero.svg
+++ b/assets/counterproof-hero.svg
@@ -3,7 +3,7 @@
   <rect x="30" y="30" width="1220" height="560" rx="22" fill="#111210"/>
   <text x="72" y="92" fill="#9ea098" font-family="ui-monospace, SFMono-Regular, Menlo, monospace" font-size="18" letter-spacing="2">COUNTERPROOF / PR PROOF LAB</text>
   <text x="72" y="162" fill="#f5f2e8" font-family="Inter, Arial, sans-serif" font-weight="700" font-size="48">Your coding agent says it fixed the bug.</text>
-  <text x="72" y="216" fill="#f5f2e8" font-family="Inter, Arial, sans-serif" font-weight="700" font-size="48">Prove it.</text>
+  <text x="72" y="216" fill="#f5f2e8" font-family="Inter, Arial, sans-serif" font-weight="700" font-size="48">Prove the exact claim.</text>
 
   <rect x="72" y="266" width="1136" height="210" rx="14" fill="#1b1d1a" stroke="#3d403a"/>
   <text x="102" y="310" fill="#8e9188" font-family="ui-monospace, SFMono-Regular, Menlo, monospace" font-size="17">REGRESSION WITNESS</text>
@@ -20,7 +20,7 @@
 
   <rect x="832" y="316" width="322" height="96" rx="12" fill="#eeeadf"/>
   <text x="864" y="352" fill="#444740" font-family="ui-monospace, SFMono-Regular, Menlo, monospace" font-size="14" letter-spacing="1">BEFORE / AFTER</text>
-  <text x="864" y="388" fill="#111210" font-family="ui-monospace, SFMono-Regular, Menlo, monospace" font-size="25" font-weight="700">PASS ≠ PROOF</text>
+  <text x="864" y="388" fill="#111210" font-family="ui-monospace, SFMono-Regular, Menlo, monospace" font-size="25" font-weight="700">PASS ≠ CLAIM PROOF</text>
 
   <rect x="72" y="516" width="316" height="48" rx="24" fill="#f5f2e8"/>
   <text x="107" y="547" fill="#111210" font-family="Inter, Arial, sans-serif" font-size="17" font-weight="700">OPEN LIVE PROOF LAB →</text>
diff --git a/assets/social-preview.svg b/assets/social-preview.svg
index 3ae2b90..e6e9d62 100644
--- a/assets/social-preview.svg
+++ b/assets/social-preview.svg
@@ -2,9 +2,9 @@
   <rect width="1280" height="640" fill="#f3f0e7"/>
   <rect x="32" y="32" width="1216" height="576" rx="28" fill="#111210"/>
   <text x="76" y="92" fill="#8f928a" font-family="ui-monospace,SFMono-Regular,Menlo,monospace" font-size="16" letter-spacing="3">COUNTERPROOF</text>
-  <text x="76" y="176" fill="#f5f2e8" font-family="Inter,Arial,sans-serif" font-size="52" font-weight="700">Your coding agent says</text>
-  <text x="76" y="236" fill="#f5f2e8" font-family="Inter,Arial,sans-serif" font-size="52" font-weight="700">it fixed the bug. Prove it.</text>
-  <text x="76" y="286" fill="#a3a69e" font-family="Inter,Arial,sans-serif" font-size="20">Behavioral proof for agent-generated pull requests.</text>
+  <text x="76" y="176" fill="#f5f2e8" font-family="Inter,Arial,sans-serif" font-size="52" font-weight="700">Your coding agent says it fixed the bug.</text>
+  <text x="76" y="236" fill="#f5f2e8" font-family="Inter,Arial,sans-serif" font-size="52" font-weight="700">Prove the exact claim.</text>
+  <text x="76" y="286" fill="#a3a69e" font-family="Inter,Arial,sans-serif" font-size="20">Same evidence · BASE vs HEAD · scope stays explicit.</text>
 
   <rect x="76" y="338" width="1128" height="174" rx="16" fill="#1b1d1a" stroke="#3d403a"/>
   <text x="108" y="378" fill="#85887f" font-family="ui-monospace,SFMono-Regular,Menlo,monospace" font-size="15">PR HEAD + CHANGED TESTS</text>
diff --git a/site/app.js b/site/app.js
index 4d334d6..697670e 100644
--- a/site/app.js
+++ b/site/app.js
@@ -20,11 +20,11 @@ const witnessScenarios = {
     baseNote: "pre-change behavior reproduced",
     verdict: "WITNESSED",
     integrity: "proof integrity · CLEAN",
-    proofStatus: "VERIFIED",
-    proofReady: "proof ready · TRUE",
+    proofStatus: "SUBMITTED JUDGE",
+    proofReady: "product oracle · UNVERIFIED",
     proofClass: "verified",
     explanation: "The changed test passes on the PR, fails on the old code, and the evidence surface is clean.",
-    takeaway: "The configured Counterproof evidence contract is satisfied."
+    takeaway: "The regression is witnessed under the submitted judge; broader product correctness remains unproven."
   },
   "weak-test": {
     head: "PASS",
@@ -34,7 +34,7 @@ const witnessScenarios = {
     verdict: "NOT WITNESSED",
     integrity: "proof integrity · CLEAN",
     proofStatus: "UNPROVEN",
-    proofReady: "proof ready · FALSE",
+    proofReady: "submitted judge · INSUFFICIENT",
     proofClass: "unproven",
     explanation: "Green on HEAD is not enough when the same test was already green on BASE.",
     takeaway: "The configured evidence does not prove the regression."
@@ -47,10 +47,10 @@ const witnessScenarios = {
     verdict: "WITNESSED",
     integrity: "proof integrity · CI/TEST SURFACE CHANGED",
     proofStatus: "REVIEW REQUIRED",
-    proofReady: "proof ready · FALSE",
+    proofReady: "judge changed · REVIEW",
     proofClass: "review",
     explanation: "The regression witness exists, but the PR also changed how evidence is produced.",
-    takeaway: "The witness survives, but the unified proof is blocked until the judge is reviewed."
+    takeaway: "The witness survives, but changed evidence-producing machinery requires reviewer inspection before any broader claim."
   },
   "suite-delta": {
     head: "PASS",
@@ -60,7 +60,7 @@ const witnessScenarios = {
     verdict: "SUITE DELTA",
     integrity: "proof integrity · CLEAN",
     proofStatus: "SUITE DELTA",
-    proofReady: "proof ready · FALSE",
+    proofReady: "exact witness · UNPROVEN",
     proofClass: "suite",
     explanation: "The whole suite distinguishes HEAD from BASE, but the changed test was not isolated.",
     takeaway: "Useful evidence, but deliberately weaker than an exact Regression Witness."
diff --git a/site/index.html b/site/index.html
index 4571908..fd12f19 100644
--- a/site/index.html
+++ b/site/index.html
@@ -3,8 +3,8 @@
 <head>
   <meta charset="utf-8">
   <meta name="viewport" content="width=device-width,initial-scale=1">
-  <meta name="description" content="Counterproof proves agent PR fixes by replaying changed tests against pre-change code, then protects the evidence from CI/test tampering.">
-  <title>Counterproof · Proof Lab</title>
+  <meta name="description" content="CounterProof replays the same PR evidence on BASE and HEAD, keeps claim scope explicit, and surfaces when the test or CI judge changed.">
+  <title>CounterProof · Proof Lab</title>
   <link rel="stylesheet" href="styles.css">
 </head>
 <body>
@@ -18,9 +18,9 @@
     <section class="opening">
       <div class="opening-index">01—</div>
       <div>
-        <p class="overline">PROOF-OF-FIX FOR AGENT PULL REQUESTS</p>
-        <h1>Green says it passes now.<br><em>Prove it failed before the fix.</em></h1>
-        <p class="opening-copy">Counterproof replays the PR’s changed tests against pre-change code, then checks whether the same PR weakened the tests or CI that produced the evidence.</p>
+        <p class="overline">EVIDENCE REPLAY FOR AI-ASSISTED PULL REQUESTS</p>
+        <h1>Green says it passes now.<br><em>Prove the exact claim.</em></h1>
+        <p class="opening-copy">CounterProof replays the same submitted evidence against pre-change code, then checks whether the PR changed the test or CI judge. It tells you what the evidence establishes — not whether the PR should merge.</p>
       </div>
     </section>
 
@@ -52,9 +52,9 @@ <h1>Green says it passes now.<br><em>Prove it failed before the fix.</em></h1>
           <strong id="witnessVerdict">WITNESSED</strong>
           <small id="witnessIntegrity">proof integrity · CLEAN</small>
           <div id="witnessUnified" class="witness-unified verified">
-            <span>UNIFIED PROOF</span>
-            <b id="witnessProofStatus">VERIFIED</b>
-            <small id="witnessProofReady">proof ready · TRUE</small>
+            <span>EVIDENCE SCOPE</span>
+            <b id="witnessProofStatus">SUBMITTED JUDGE</b>
+            <small id="witnessProofReady">product oracle · UNVERIFIED</small>
           </div>
         </div>
       </div>
@@ -234,7 +234,7 @@ <h2>What is real today,<br>and what is still a research target.</h2>
   <div id="toast" class="toast" aria-live="polite"></div>
 
   <footer>
-    <span>COUNTERPROOF / SKILLFACTORY</span>
+    <span>COUNTERPROOF / PROOF LAB</span>
     <span>Agents will learn. The question is whether you can review what they learned.</span>
   </footer>
 
diff --git a/site/standalone.html b/site/standalone.html
index d97fc2f..1303d44 100644
--- a/site/standalone.html
+++ b/site/standalone.html
@@ -3,8 +3,8 @@
 <head>
   <meta charset="utf-8">
   <meta name="viewport" content="width=device-width,initial-scale=1">
-  <meta name="description" content="Counterproof proves agent PR fixes by replaying changed tests against pre-change code, then protects the evidence from CI/test tampering.">
-  <title>Counterproof · Proof Lab</title>
+  <meta name="description" content="CounterProof replays the same PR evidence on BASE and HEAD, keeps claim scope explicit, and surfaces when the test or CI judge changed.">
+  <title>CounterProof · Proof Lab</title>
   <style>@import url('https://fonts.googleapis.com/css2?family=DM+Mono:wght@300;400;500&family=Manrope:wght@400;500;600;700&family=Newsreader:ital,opsz,wght@1,6..72,500&display=swap');
 :root{--paper:#f2efe8;--ink:#171814;--muted:#77796f;--rule:#cbc7bc;--red:#d74c35;--blue:#3f63a8;--green:#386c49;--yellow:#e2d46d;--white:#fbfaf6}
 *{box-sizing:border-box}html{background:var(--paper)}body{margin:0;color:var(--ink);background:var(--paper);font:15px/1.55 Manrope,system-ui,sans-serif}button{font:inherit}button:focus-visible,a:focus-visible{outline:2px solid var(--blue);outline-offset:3px}.mast{height:58px;border-bottom:1px solid var(--ink);display:grid;grid-template-columns:1fr auto 1fr;align-items:center;padding:0 28px;font:11px DM Mono,monospace;letter-spacing:.05em}.wordmark{color:var(--ink);font-weight:500;text-decoration:none;font-size:14px}.mast-center{text-align:center}.mast-right{justify-self:end;display:flex;gap:20px;color:var(--muted)}.mast-right a{color:var(--ink);text-decoration:none}main{max-width:1320px;margin:auto;padding:0 28px}.opening,.capability-truth{display:grid;grid-template-columns:90px 1fr;padding:86px 0 58px}.opening-index{font:12px DM Mono,monospace;color:var(--muted);padding-top:9px}.overline{font:10px DM Mono,monospace;letter-spacing:.13em;color:var(--muted);margin:0}.opening h1{font-size:clamp(42px,6.1vw,82px);line-height:.95;letter-spacing:-.064em;margin:14px 0 24px;max-width:1050px}.opening h1 em,.capability-truth h2 em{font-family:Newsreader,serif;font-weight:500}.opening-copy{font-size:18px;max-width:690px;color:#55584f}
@@ -34,9 +34,9 @@
     <section class="opening">
       <div class="opening-index">01—</div>
       <div>
-        <p class="overline">PROOF-OF-FIX FOR AGENT PULL REQUESTS</p>
-        <h1>Green says it passes now.<br><em>Prove it failed before the fix.</em></h1>
-        <p class="opening-copy">Counterproof replays the PR’s changed tests against pre-change code, then checks whether the same PR weakened the tests or CI that produced the evidence.</p>
+        <p class="overline">EVIDENCE REPLAY FOR AI-ASSISTED PULL REQUESTS</p>
+        <h1>Green says it passes now.<br><em>Prove the exact claim.</em></h1>
+        <p class="opening-copy">CounterProof replays the same submitted evidence against pre-change code, then checks whether the PR changed the test or CI judge. It tells you what the evidence establishes — not whether the PR should merge.</p>
       </div>
     </section>
 
@@ -68,9 +68,9 @@ <h1>Green says it passes now.<br><em>Prove it failed before the fix.</em></h1>
           <strong id="witnessVerdict">WITNESSED</strong>
           <small id="witnessIntegrity">proof integrity · CLEAN</small>
           <div id="witnessUnified" class="witness-unified verified">
-            <span>UNIFIED PROOF</span>
-            <b id="witnessProofStatus">VERIFIED</b>
-            <small id="witnessProofReady">proof ready · TRUE</small>
+            <span>EVIDENCE SCOPE</span>
+            <b id="witnessProofStatus">SUBMITTED JUDGE</b>
+            <small id="witnessProofReady">product oracle · UNVERIFIED</small>
           </div>
         </div>
       </div>
@@ -250,7 +250,7 @@ <h2>What is real today,<br>and what is still a research target.</h2>
   <div id="toast" class="toast" aria-live="polite"></div>
 
   <footer>
-    <span>COUNTERPROOF / SKILLFACTORY</span>
+    <span>COUNTERPROOF / PROOF LAB</span>
     <span>Agents will learn. The question is whether you can review what they learned.</span>
   </footer>
 
@@ -922,7 +922,7 @@ <h2>What is real today,<br>and what is still a research target.</h2>
       "id": "proof-integrity-guard",
       "name": "Proof Integrity Guard for changed tests and CI",
       "status": "tested",
-      "evidence": "Counterproof deterministically flags deleted tests, added skip/xfail markers, shell failure masking, continue-on-error, removed pull-request workflow triggers, and other evidence-surface changes.",
+      "evidence": "CounterProof deterministically flags deleted tests, added skip/xfail markers, shell failure masking, continue-on-error, removed pull-request workflow triggers, and other evidence-surface changes.",
       "limitation": "A finding means the judging machinery changed and needs review; it is not proof that the change is malicious or incorrect."
     },
     {
@@ -930,7 +930,7 @@ <h2>What is real today,<br>and what is still a research target.</h2>
       "name": "Install-level real-Git doctor self-test",
       "status": "tested",
       "evidence": "counterproof doctor creates a temporary Git repository and exercises structured-result parsing, Regression Witness, and Proof Integrity without an external service.",
-      "limitation": "Doctor validates Counterproof mechanics, not the consuming repository's full dependency graph or domain semantics."
+      "limitation": "Doctor validates CounterProof mechanics, not the consuming repository's full dependency graph or domain semantics."
     },
     {
       "id": "github-onboarding",
@@ -943,14 +943,14 @@ <h2>What is real today,<br>and what is still a research target.</h2>
       "id": "regression-witness",
       "name": "Regression Witness for agent pull requests",
       "status": "tested",
-      "evidence": "In precise mode, Counterproof runs the exact changed tests on PR head and detached base code with changed test support overlaid. Full-suite commands are reported separately as SUITE DELTA.",
+      "evidence": "In precise mode, CounterProof runs the exact changed tests on PR head and detached base code with changed test support overlaid. Full-suite commands are reported separately as SUITE DELTA.",
       "limitation": "Suite-level evidence is deliberately weaker and cannot satisfy --require-witness; repository-specific dependency setup still belongs to the consuming project."
     },
     {
       "id": "proof-receipt",
       "name": "Reproducible Proof Receipt with source fingerprints",
       "status": "tested",
-      "evidence": "Counterproof can hash the exact trace and experiment manifest, record observed/expected signatures and selection results, then verify later that the proof still points to unchanged inputs.",
+      "evidence": "CounterProof can hash the exact trace and experiment manifest, record observed/expected signatures and selection results, then verify later that the proof still points to unchanged inputs.",
       "limitation": "The receipt fingerprints source artifacts; it does not yet capture a full container image, dependency lock or external service state."
     },
     {
@@ -979,7 +979,7 @@ <h2>What is real today,<br>and what is still a research target.</h2>
       "name": "Multiple mutation surfaces",
       "status": "partial",
       "evidence": "The data model accepts skill, prompt, policy, router, memory, tool, and eval surfaces.",
-      "limitation": "Counterproof does not yet apply each mutation surface to a live agent automatically."
+      "limitation": "CounterProof does not yet apply each mutation surface to a live agent automatically."
     },
     {
       "id": "evidence-semantics",
@@ -1021,7 +1021,7 @@ <h2>What is real today,<br>and what is still a research target.</h2>
       "name": "Pre-registered PASS / FAIL predictions for intervention cases",
       "status": "tested",
       "evidence": "Experiment manifests can register expected outcomes before execution; contradicted or partial predictions are tracked separately from runtime pass/fail and can block automatic selection.",
-      "limitation": "Prediction quality still depends on the experiment author or adapter; Counterproof does not yet learn calibrated causal predictions."
+      "limitation": "Prediction quality still depends on the experiment author or adapter; CounterProof does not yet learn calibrated causal predictions."
     },
     {
       "id": "active-discrimination",
@@ -1069,7 +1069,7 @@ <h2>What is real today,<br>and what is still a research target.</h2>
       "id": "next-probe-planner",
       "name": "Next discriminating probe planning for ambiguous survivors",
       "status": "tested",
-      "evidence": "When survivor behavior signatures are identical, Counterproof emits a pairwise plan stating what to hold fixed, what levers to vary, competing predictions, and a falsification rule.",
+      "evidence": "When survivor behavior signatures are identical, CounterProof emits a pairwise plan stating what to hold fixed, what levers to vary, competing predictions, and a falsification rule.",
       "limitation": "The planner produces experiment-design guidance, not an automatically instantiated executable domain test."
     },
     {
@@ -1105,7 +1105,7 @@ <h2>What is real today,<br>and what is still a research target.</h2>
       "name": "Automatic Evolution PR open / merge / revert",
       "status": "planned",
       "evidence": "Roadmap only.",
-      "limitation": "No GitHub automation is wired into the Counterproof runtime yet."
+      "limitation": "No GitHub automation is wired into the CounterProof runtime yet."
     },
     {
       "id": "online-rollout",
@@ -1221,11 +1221,11 @@ <h2>What is real today,<br>and what is still a research target.</h2>
     baseNote: "pre-change behavior reproduced",
     verdict: "WITNESSED",
     integrity: "proof integrity · CLEAN",
-    proofStatus: "VERIFIED",
-    proofReady: "proof ready · TRUE",
+    proofStatus: "SUBMITTED JUDGE",
+    proofReady: "product oracle · UNVERIFIED",
     proofClass: "verified",
     explanation: "The changed test passes on the PR, fails on the old code, and the evidence surface is clean.",
-    takeaway: "The configured Counterproof evidence contract is satisfied."
+    takeaway: "The regression is witnessed under the submitted judge; broader product correctness remains unproven."
   },
   "weak-test": {
     head: "PASS",
@@ -1235,7 +1235,7 @@ <h2>What is real today,<br>and what is still a research target.</h2>
     verdict: "NOT WITNESSED",
     integrity: "proof integrity · CLEAN",
     proofStatus: "UNPROVEN",
-    proofReady: "proof ready · FALSE",
+    proofReady: "submitted judge · INSUFFICIENT",
     proofClass: "unproven",
     explanation: "Green on HEAD is not enough when the same test was already green on BASE.",
     takeaway: "The configured evidence does not prove the regression."
@@ -1248,10 +1248,10 @@ <h2>What is real today,<br>and what is still a research target.</h2>
     verdict: "WITNESSED",
     integrity: "proof integrity · CI/TEST SURFACE CHANGED",
     proofStatus: "REVIEW REQUIRED",
-    proofReady: "proof ready · FALSE",
+    proofReady: "judge changed · REVIEW",
     proofClass: "review",
     explanation: "The regression witness exists, but the PR also changed how evidence is produced.",
-    takeaway: "The witness survives, but the unified proof is blocked until the judge is reviewed."
+    takeaway: "The witness survives, but changed evidence-producing machinery requires reviewer inspection before any broader claim."
   },
   "suite-delta": {
     head: "PASS",
@@ -1261,7 +1261,7 @@ <h2>What is real today,<br>and what is still a research target.</h2>
     verdict: "SUITE DELTA",
     integrity: "proof integrity · CLEAN",
     proofStatus: "SUITE DELTA",
-    proofReady: "proof ready · FALSE",
+    proofReady: "exact witness · UNPROVEN",
     proofClass: "suite",
     explanation: "The whole suite distinguishes HEAD from BASE, but the changed test was not isolated.",
     takeaway: "Useful evidence, but deliberately weaker than an exact Regression Witness."