diff --git a/src/queue-intelligence.ts b/src/queue-intelligence.ts
index b2b28b2959..fd639c5181 100644
--- a/src/queue-intelligence.ts
+++ b/src/queue-intelligence.ts
@@ -44,6 +44,10 @@ export const FORBIDDEN_PUBLIC_COMMENT_WORDS = [
   "wallet",
   "hotkey",
   "raw trust score",
+  "trust score",
+  "coldkey",
+  "seed phrase",
+  "mnemonic",
   "payout",
   "reward estimate",
   "estimated rewards",
@@ -52,7 +56,14 @@ export const FORBIDDEN_PUBLIC_COMMENT_WORDS = [
   "reward",
   "farming",
   "private reviewability",
+  "reviewability internals",
+  "reviewability",
+  "private scoreability",
+  "scoreability",
+  "score preview",
+  "estimated score",
   "public score estimate",
+  "score estimate",
   "private rankings",
   "private ranking",
   "rankings",
diff --git a/src/services/ai-summaries.ts b/src/services/ai-summaries.ts
index 1b476232c3..db8b0221fd 100644
--- a/src/services/ai-summaries.ts
+++ b/src/services/ai-summaries.ts
@@ -8,10 +8,10 @@ const PR_INTELLIGENCE_MARKER = "<!-- gittensory-pr-intelligence -->";
 type AiSummaryVisibility = "private" | "public";
 
 const PRIVATE_CONTEXT_PATTERN =
-  /\b(wallets?|hotkeys?|coldkeys?|seed phrases?|mnemonics?|raw trust scores?|trust scores?|private reviewability|private scoreability|public score estimates?)\b/gi;
+  /\b(wallets?|hotkeys?|coldkeys?|seed phrases?|mnemonics?|raw trust scores?|trust scores?|scoreability|reviewability(?: internals?)?|private reviewability|private scoreability|public score estimates?)\b/gi;
 const PRIVATE_OUTCOME_PATTERN = /\b(payouts?|farming|rewards?|reward estimates?|reward optimization)\b/gi;
 const PUBLIC_FORBIDDEN_TEXT_PATTERN =
-  /\b(wallets?|hotkeys?|coldkeys?|seed phrases?|mnemonics?|raw trust scores?|trust scores?|estimated scores?|score estimates?|public score estimates?|estimated rewards?|rewards?|reward estimates?|payouts?|farming|private reviewability|private scoreability|private rankings?|rankings?|reward optimization)\b/i;
+  /\b(wallets?|hotkeys?|coldkeys?|seed phrases?|mnemonics?|raw trust scores?|trust scores?|estimated scores?|score estimates?|scoreability|score preview|public score estimates?|estimated rewards?|rewards?|reward estimates?|payouts?|farming|reviewability(?: internals?)?|private reviewability|private scoreability|private rankings?|rankings?|reward optimization)\b/i;
 
 export type AiSummaryResult =
   | { status: "disabled"; reason: string }
@@ -55,7 +55,7 @@ export async function summarizeAgentBundleWithAi(env: Env, bundle: AgentRunBundl
           role: "system",
           content:
             visibility === "public"
-              ? "Summarize deterministic Gittensory signals for a public GitHub comment. Do not mention rewards, rankings, payouts, wallets, hotkeys, raw trust scores, or private reviewability."
+              ? "Summarize deterministic Gittensory signals for a public GitHub comment. Do not mention rewards, rankings, payouts, wallets, hotkeys, raw trust scores, scoreability, or reviewability."
               : "Summarize deterministic Gittensory signals for an authenticated MCP/API user. Be concise and preserve scoreability blockers and next actions.",
         },
         { role: "user", content: prompt },
@@ -175,7 +175,8 @@ function sanitizeAiText(value: string, visibility: AiSummaryVisibility): string
 function containsPublicForbiddenText(value: string): boolean {
   // Route every public AI output through the canonical public/private sanitizer (issue #151).
   // `sanitizePublicComment` throws on any forbidden public term (wallet, hotkey, raw trust score,
-  // payout, reward language, farming, private reviewability, public score estimate, or ranking language).
+  // scoreability/reviewability terms, payout, reward language, farming, public score estimate,
+  // or ranking language).
   try {
     sanitizePublicComment(value);
   } catch {
@@ -327,7 +328,7 @@ export async function rewritePublicPrIntelligenceComment(
     bundle: args.bundle,
     fallbackText: args.deterministicBody,
     instructions:
-      "Rewrite this deterministic Gittensory PR signal bundle as a short, friendly public GitHub comment with 3-5 bullet points. Only restate the facts provided. Never mention rewards, rankings, payouts, wallets, hotkeys, raw or estimated trust scores, score estimates, farming, or private reviewability, and never claim a guaranteed outcome.",
+      "Rewrite this deterministic Gittensory PR signal bundle as a short, friendly public GitHub comment with 3-5 bullet points. Only restate the facts provided. Never mention rewards, rankings, payouts, wallets, hotkeys, raw or estimated trust scores, score estimates, scoreability, reviewability, or farming, and never claim a guaranteed outcome.",
     actor: args.actor,
     route: args.route,
   });
diff --git a/test/unit/ai-summaries.test.ts b/test/unit/ai-summaries.test.ts
index 04c9d3bea2..a48c270b1b 100644
--- a/test/unit/ai-summaries.test.ts
+++ b/test/unit/ai-summaries.test.ts
@@ -10,7 +10,7 @@ import { FORBIDDEN_PUBLIC_COMMENT_WORDS } from "../../src/queue-intelligence";
 import { createTestEnv } from "../helpers/d1";
 
 const PUBLIC_FORBIDDEN_TEXT =
-  /\b(wallets?|hotkeys?|raw trust scores?|trust scores?|payouts?|estimated rewards?|rewards?|reward estimates?|farming|private reviewability|private scoreability|private rankings?|rankings?|public score estimates?)\b/i;
+  /\b(wallets?|hotkeys?|raw trust scores?|trust scores?|payouts?|estimated rewards?|rewards?|reward estimates?|farming|scoreability|reviewability(?: internals?)?|private reviewability|private scoreability|private rankings?|rankings?|public score estimates?)\b/i;
 type AiRunRequest = { messages: Array<{ role: string; content: string }> };
 
 describe("Workers AI summaries", () => {
@@ -152,6 +152,9 @@ describe("Workers AI summaries", () => {
     "rewards",
     "reward",
     "farming",
+    "scoreability",
+    "reviewability",
+    "reviewability internals",
     "private reviewability",
     "private scoreability",
     "public score estimate",
@@ -350,12 +353,14 @@ describe("optional deterministic-summary rewrite layer", () => {
     }
   });
 
-  it("rejects reward and ranking variants in public AI rewrites", async () => {
+  it("rejects reward, ranking, scoreability, and reviewability variants in public AI rewrites", async () => {
     const unsafeOutputs = [
       "The estimated reward is high.",
       "Rewards look likely for this PR.",
+      "This PR may improve contributor rewards and scoreability.",
       "The private ranking looks strong.",
       "This ranking should improve.",
+      "Reviewability internals look healthy.",
     ];
 
     for (const output of unsafeOutputs) {