diff --git a/.jules/bolt.md b/.jules/bolt.md
index 49b1916..8c9009f 100644
--- a/.jules/bolt.md
+++ b/.jules/bolt.md
@@ -47,3 +47,7 @@
 ## 2026-07-02 - Remove `re.search` fast-path pre-check
 **Learning:** Python's `re.finditer` evaluates lazily by allocating a lightweight C-level `ScannerObject`. Using `re.search` as a fast-path pre-check before `re.finditer` is an anti-pattern that addresses a non-existent bottleneck and degrades performance for matched paths by evaluating the regex twice.
 **Action:** Do not use `re.search` before `re.finditer` for optimization purposes.
+
+## 2024-07-03 - Optimize set deduplication with dict.fromkeys
+**Learning:** In Python, iterating over a list to maintain insertion-order uniqueness (e.g. `if ref not in seen: seen.append(ref)`) requires O(N) lookup time, resulting in O(N^2) complexity overall.
+**Action:** Use `dict.fromkeys(iterator)` which leverages Python's highly optimized, insertion-ordered C dictionary implementation to achieve fast deduplication in O(N) total time.
diff --git a/appguardrail_core/rules.py b/appguardrail_core/rules.py
index 87180a8..8a3074a 100644
--- a/appguardrail_core/rules.py
+++ b/appguardrail_core/rules.py
@@ -90,6 +90,12 @@ def as_dict(self) -> dict[str, Any]:
 
 def extract_public_references(message: str) -> tuple[str, ...]:
     """Extract OWASP, CWE, and CVE references already embedded in rule copy."""
+    # ⚡ Bolt: Fast substring pre-filter to bypass expensive regex evaluation
+    # and match object allocations when the message definitely does not
+    # contain a public reference token.
+    if message and not ("CWE-" in message or "OWASP " in message or "CVE-" in message):
+        return ()
+
     seen: list[str] = []
     for match in REFERENCE_RE.finditer(message or ""):
         reference = " ".join(match.group(1).split())
@@ -140,9 +146,4 @@ def validate_rule_metadata(metadata: RuleMetadata | dict[str, Any]) -> list[str]
 
 
 def _merge_references(*groups: tuple[str, ...]) -> tuple[str, ...]:
-    merged: list[str] = []
-    for group in groups:
-        for reference in group:
-            if reference and reference not in merged:
-                merged.append(reference)
-    return tuple(merged)
+    return tuple(dict.fromkeys(ref for group in groups for ref in group if ref))