diff --git a/src/scoring/preview.ts b/src/scoring/preview.ts
index 7e495c32f2..444d7b0ad8 100644
--- a/src/scoring/preview.ts
+++ b/src/scoring/preview.ts
@@ -756,13 +756,55 @@ function deltaExplanationFor(core: ScoreCore, blockedBy: ScoreGateBlocker[]): st
 }
 
 function selectLabelMultiplier(labels: string[], multipliers: Record<string, number>, fallback: number): number {
-  const normalized = new Set(labels.map((label) => label.toLowerCase()));
-  const matched = Object.entries(multipliers).flatMap(([label, multiplier]) =>
-    normalized.has(label.toLowerCase()) ? [multiplier] : [],
-  );
+  const normalized = labels.map((label) => label.toLowerCase());
+  const matched = Object.entries(multipliers).flatMap(([pattern, multiplier]) => {
+    const matcher = labelPatternToRegExp(pattern.toLowerCase());
+    return normalized.some((label) => matcher.test(label)) ? [multiplier] : [];
+  });
   return matched.length > 0 ? Math.max(...matched) : fallback || 1;
 }
 
+// Upstream resolves label multipliers by matching each configured key as a Python `fnmatch` GLOB, not a
+// literal string: `fnmatch(label.lower(), pattern.lower())` in
+// gittensor/validator/oss_contributions/label_resolution.py, so a repo can configure `type:*`, `kind/*`, or
+// `priority:?` and have it match `type:bug-fix`, `kind/bug`, `priority:1` (#1244-class scoring parity). The
+// preview previously did exact equality, so it silently scored every wildcard-configured trusted label at the
+// neutral default — under-/over-estimating the score for any repo using glob keys. Translate one fnmatch
+// pattern to an anchored, case-insensitive RegExp. fnmatch semantics differ from the path-glob in
+// change-guardrail.ts (there `*` stops at `/` and `?` is literal): labels are flat strings, so `*` matches any
+// run, `?` any single character, and `[seq]`/`[!seq]` a character class. Literal keys are unaffected — for a
+// pattern with no glob metacharacter the RegExp is an exact match, so existing configs score identically.
+function labelPatternToRegExp(pattern: string): RegExp {
+  let regex = "";
+  let i = 0;
+  while (i < pattern.length) {
+    const char = pattern.charAt(i);
+    i += 1;
+    if (char === "*") {
+      regex += ".*";
+    } else if (char === "?") {
+      regex += ".";
+    } else if (char === "[") {
+      const close = pattern.indexOf("]", i);
+      if (close === -1) {
+        // No closing bracket: fnmatch treats the `[` as a literal character.
+        regex += "\\[";
+      } else {
+        let body = pattern.slice(i, close).replace(/\\/g, "\\\\");
+        // `[!seq]` is fnmatch's negated class; RegExp spells negation as `[^seq]`.
+        if (body.startsWith("!")) body = `^${body.slice(1)}`;
+        regex += `[${body}]`;
+        i = close + 1;
+      }
+    } else if (/[.+^${}()|\]\\]/.test(char)) {
+      regex += `\\${char}`;
+    } else {
+      regex += char;
+    }
+  }
+  return new RegExp(`^${regex}$`, "i");
+}
+
 function decideLinkedIssueMultiplier(
   mode: "none" | "standard" | "maintainer",
   context: LinkedIssueMultiplierContext | undefined,
diff --git a/test/unit/scoring.test.ts b/test/unit/scoring.test.ts
index d25ecc7bf1..4df2e1b1b3 100644
--- a/test/unit/scoring.test.ts
+++ b/test/unit/scoring.test.ts
@@ -803,6 +803,44 @@ NOVELTY_BONUS_SCALAR = 3
     );
   });
 
+  it("matches configured label keys as fnmatch globs, mirroring the upstream validator", () => {
+    const baseInput: ScorePreviewInput = {
+      repoFullName: repo.fullName,
+      sourceTokenScore: 60,
+      totalTokenScore: 90,
+      sourceLines: 50,
+      openPrCount: 0,
+      credibility: 1,
+      linkedIssueMode: "none",
+    };
+    const labelMultiplierFor = (labelMultipliers: Record<string, number>, labels: string[], defaultLabelMultiplier = 1): number =>
+      buildScorePreview({
+        repo: { ...repo, registryConfig: { ...repo.registryConfig!, defaultLabelMultiplier, labelMultipliers } },
+        snapshot,
+        input: { ...baseInput, labels },
+      }).scoreEstimate.labelMultiplier;
+
+    // `*` spans any run of characters (including `/` and `:` — labels are flat strings, not paths).
+    expect(labelMultiplierFor({ "kind/*": 1.5 }, ["kind/bug"])).toBe(1.5);
+    expect(labelMultiplierFor({ "type:*": 1.1 }, ["type:bug-fix"])).toBe(1.1);
+    // `?` matches exactly one character: it matches `priority:1` but not the two-digit `priority:10`.
+    expect(labelMultiplierFor({ "priority:?": 2 }, ["priority:1"])).toBe(2);
+    expect(labelMultiplierFor({ "priority:?": 2 }, ["priority:10"])).toBe(1);
+    // `[seq]` / `[!seq]` character classes.
+    expect(labelMultiplierFor({ "[bf]ug": 1.4 }, ["bug"])).toBe(1.4);
+    expect(labelMultiplierFor({ "[!x]ug": 1.3 }, ["bug"])).toBe(1.3);
+    // A `[` with no closing bracket is a literal, not a class.
+    expect(labelMultiplierFor({ "a[b": 0.7 }, ["a[b"])).toBe(0.7);
+    // Regex metacharacters in a literal key stay literal: `.` matches only a dot, not any char.
+    expect(labelMultiplierFor({ "v1.0": 1.1 }, ["v1.0"])).toBe(1.1);
+    expect(labelMultiplierFor({ "v1.0": 1.1 }, ["v1x0"])).toBe(1);
+    // When several patterns match, the highest multiplier wins (mirrors upstream `max(...)`).
+    expect(labelMultiplierFor({ "kind/*": 1.1, "*/bug": 1.6 }, ["kind/bug"])).toBe(1.6);
+    // Literal keys are unchanged — exact match, parity-preserving for every existing config.
+    expect(labelMultiplierFor({ bug: 1.2 }, ["bug"])).toBe(1.2);
+    expect(labelMultiplierFor({ bug: 1.2 }, ["feature"])).toBe(1);
+  });
+
   it("gates linked-issue assumptions with branch eligibility evidence", () => {
     const baseInput = {
       repoFullName: repo.fullName,