diff --git a/src/components/nostr/RichText.tsx b/src/components/nostr/RichText.tsx
index d5487a3..72ce91e 100644
--- a/src/components/nostr/RichText.tsx
+++ b/src/components/nostr/RichText.tsx
@@ -12,6 +12,7 @@ import { Nip } from "./RichText/Nip";
 import { Relay } from "./RichText/Relay";
 import { nipReferences } from "@/lib/nip-transformer";
 import { relayReferences } from "@/lib/relay-transformer";
+import { nostrMentionReferences } from "@/lib/nostr-mention-transformer";
 import type { NostrEvent } from "@/types/nostr";
 import type { Root } from "applesauce-content/nast";
 
@@ -23,8 +24,11 @@ const { useRenderedContent } = Hooks;
 // Custom cache key for our extended transformers
 const GrimoireContentSymbol = Symbol.for("grimoire-content");
 
-// Default transformers including our custom NIP and relay transformers
+// Default transformers including our custom NIP, relay, and mention transformers
+// IMPORTANT: nostrMentionReferences must come BEFORE textNoteTransformers
+// to handle edge cases where mentions are immediately followed by text without whitespace
 export const defaultTransformers = [
+  nostrMentionReferences,
   ...textNoteTransformers,
   nipReferences,
   relayReferences,
diff --git a/src/lib/nostr-mention-transformer.test.ts b/src/lib/nostr-mention-transformer.test.ts
new file mode 100644
index 0000000..925756f
--- /dev/null
+++ b/src/lib/nostr-mention-transformer.test.ts
@@ -0,0 +1,207 @@
+import { describe, it, expect } from "vitest";
+import { nostrMentionReferences } from "./nostr-mention-transformer";
+import type { Root } from "applesauce-content/nast";
+
+/**
+ * Helper to create a simple text tree for testing
+ */
+function createTextTree(content: string): Root {
+  return {
+    type: "root",
+    children: [
+      {
+        type: "paragraph",
+        children: [
+          {
+            type: "text",
+            value: content,
+          },
+        ],
+      } as any, // Type assertion needed for test setup
+    ],
+  };
+}
+
+/**
+ * Helper to extract text and mention nodes from a transformed tree
+ */
+function extractNodes(
+  tree: Root,
+): Array<{ type: string; value?: string; decoded?: any }> {
+  const results: Array<{ type: string; value?: string; decoded?: any }> = [];
+
+  for (const child of tree.children as any[]) {
+    if (child.type === "paragraph" && "children" in child) {
+      for (const node of child.children) {
+        if (node.type === "text" && "value" in node) {
+          results.push({ type: "text", value: node.value as string });
+        } else if (node.type === "mention" && "decoded" in node) {
+          results.push({ type: "mention", decoded: node.decoded });
+        }
+      }
+    }
+  }
+
+  return results;
+}
+
+describe("nostrMentionReferences", () => {
+  describe("npub mentions", () => {
+    it("should parse npub mention with space after", () => {
+      const tree = createTextTree(
+        "Hello nostr:npub107jk7htfv243u0x5ynn43scq9wrxtaasmrwwa8lfu2ydwag6cx2quqncxg world",
+      );
+      const transformer = nostrMentionReferences();
+      transformer(tree);
+
+      const nodes = extractNodes(tree);
+      expect(nodes).toHaveLength(3);
+      expect(nodes[0]).toEqual({ type: "text", value: "Hello " });
+      expect(nodes[1].type).toBe("mention");
+      expect(nodes[1].decoded?.type).toBe("npub");
+      expect(nodes[2]).toEqual({ type: "text", value: " world" });
+    });
+
+    it("should parse npub mention immediately followed by text without space", () => {
+      // This is the bug case from the issue - npub followed by "how" without space
+      const tree = createTextTree(
+        "nostr:npub107jk7htfv243u0x5ynn43scq9wrxtaasmrwwa8lfu2ydwag6cx2quqncxghow does the COUNT work?",
+      );
+      const transformer = nostrMentionReferences();
+      transformer(tree);
+
+      const nodes = extractNodes(tree);
+      expect(nodes).toHaveLength(2);
+      expect(nodes[0].type).toBe("mention");
+      expect(nodes[0].decoded?.type).toBe("npub");
+      // The "how does the COUNT work?" should remain as text
+      expect(nodes[1]).toEqual({
+        type: "text",
+        value: "how does the COUNT work?",
+      });
+    });
+
+    it("should parse npub mention followed by punctuation", () => {
+      const tree = createTextTree(
+        "Check out nostr:npub107jk7htfv243u0x5ynn43scq9wrxtaasmrwwa8lfu2ydwag6cx2quqncxg!",
+      );
+      const transformer = nostrMentionReferences();
+      transformer(tree);
+
+      const nodes = extractNodes(tree);
+      expect(nodes).toHaveLength(3);
+      expect(nodes[0]).toEqual({ type: "text", value: "Check out " });
+      expect(nodes[1].type).toBe("mention");
+      expect(nodes[1].decoded?.type).toBe("npub");
+      expect(nodes[2]).toEqual({ type: "text", value: "!" });
+    });
+
+    it("should not match nostr: without valid prefix", () => {
+      const tree = createTextTree("nostr:invalid123");
+      const transformer = nostrMentionReferences();
+      transformer(tree);
+
+      const nodes = extractNodes(tree);
+      expect(nodes).toHaveLength(1);
+      expect(nodes[0]).toEqual({ type: "text", value: "nostr:invalid123" });
+    });
+  });
+
+  describe("note mentions", () => {
+    it("should fallback to text for invalid note mention (wrong checksum)", () => {
+      // Note with invalid checksum will be matched by pattern but fail decode, falling back to text
+      const tree = createTextTree(
+        "See nostr:note1q9m4f9qx3p2l8zq3xz9h3whxuc2h69k2xy6x5agdchp3upynhsxqqdhcxs test",
+      );
+      const transformer = nostrMentionReferences();
+      transformer(tree);
+
+      const nodes = extractNodes(tree);
+      // Pattern matches, but decode fails, so matched text + surrounding text remain
+      expect(nodes).toHaveLength(3);
+      expect(nodes[0]).toEqual({ type: "text", value: "See " });
+      expect(nodes[1].type).toBe("text"); // Failed mention becomes text
+      expect(nodes[1].value).toContain("nostr:note1");
+      expect(nodes[2]).toEqual({ type: "text", value: " test" });
+    });
+  });
+
+  describe("nevent mentions", () => {
+    it("should not parse nevent that is too short", () => {
+      // nevent that's less than 40 chars after prefix won't match our pattern
+      const tree = createTextTree("Check nostr:nevent1short");
+      const transformer = nostrMentionReferences();
+      transformer(tree);
+
+      const nodes = extractNodes(tree);
+      // Should remain as text because nevent is too short
+      expect(nodes).toHaveLength(1);
+      expect(nodes[0]).toEqual({
+        type: "text",
+        value: "Check nostr:nevent1short",
+      });
+    });
+  });
+
+  describe("multiple mentions", () => {
+    it("should parse multiple mentions in one string", () => {
+      const tree = createTextTree(
+        "nostr:npub107jk7htfv243u0x5ynn43scq9wrxtaasmrwwa8lfu2ydwag6cx2quqncxg and nostr:npub107jk7htfv243u0x5ynn43scq9wrxtaasmrwwa8lfu2ydwag6cx2quqncxg!",
+      );
+      const transformer = nostrMentionReferences();
+      transformer(tree);
+
+      const nodes = extractNodes(tree);
+      // Should have: mention, text(" and "), mention, text("!")
+      expect(nodes.length).toBe(4);
+      expect(nodes[0].type).toBe("mention");
+      expect(nodes[1]).toEqual({ type: "text", value: " and " });
+      expect(nodes[2].type).toBe("mention");
+      expect(nodes[3]).toEqual({ type: "text", value: "!" });
+    });
+  });
+
+  describe("edge cases", () => {
+    it("should not match nostr: in middle of word", () => {
+      const tree = createTextTree(
+        "testnostr:npub107jk7htfv243u0x5ynn43scq9wrxtaasmrwwa8lfu2ydwag6cx2quqncxg",
+      );
+      const transformer = nostrMentionReferences();
+      transformer(tree);
+
+      const nodes = extractNodes(tree);
+      // Should remain as text because no word boundary before nostr:
+      expect(nodes).toHaveLength(1);
+      expect(nodes[0]).toEqual({
+        type: "text",
+        value:
+          "testnostr:npub107jk7htfv243u0x5ynn43scq9wrxtaasmrwwa8lfu2ydwag6cx2quqncxg",
+      });
+    });
+
+    it("should handle mention at start of string", () => {
+      const tree = createTextTree(
+        "nostr:npub107jk7htfv243u0x5ynn43scq9wrxtaasmrwwa8lfu2ydwag6cx2quqncxg",
+      );
+      const transformer = nostrMentionReferences();
+      transformer(tree);
+
+      const nodes = extractNodes(tree);
+      expect(nodes).toHaveLength(1);
+      expect(nodes[0].type).toBe("mention");
+    });
+
+    it("should handle mention at end of string", () => {
+      const tree = createTextTree(
+        "Check out nostr:npub107jk7htfv243u0x5ynn43scq9wrxtaasmrwwa8lfu2ydwag6cx2quqncxg",
+      );
+      const transformer = nostrMentionReferences();
+      transformer(tree);
+
+      const nodes = extractNodes(tree);
+      expect(nodes).toHaveLength(2);
+      expect(nodes[0]).toEqual({ type: "text", value: "Check out " });
+      expect(nodes[1].type).toBe("mention");
+    });
+  });
+});
diff --git a/src/lib/nostr-mention-transformer.ts b/src/lib/nostr-mention-transformer.ts
new file mode 100644
index 0000000..c4bc49a
--- /dev/null
+++ b/src/lib/nostr-mention-transformer.ts
@@ -0,0 +1,81 @@
+import { findAndReplace } from "applesauce-content/nast";
+import type { Root, Content } from "applesauce-content/nast";
+import { nip19 } from "nostr-tools";
+
+/**
+ * Custom node type for nostr mentions
+ */
+export interface NostrMentionNode {
+  type: "mention";
+  /** The decoded nip19 entity */
+  decoded?: {
+    type: string;
+    data: any;
+  };
+  /** The original encoded string (without nostr: prefix) */
+  encoded?: string;
+  /** The raw matched text (including nostr: prefix) */
+  raw: string;
+}
+
+/**
+ * Match nostr: URIs with proper word boundaries and length constraints
+ *
+ * Pattern explanation:
+ * - \bnostr: - word boundary + nostr: prefix
+ * - Two alternatives:
+ *   1. (npub1|note1)[bech32]{58} - fixed-length identifiers (32-byte pubkey/event ID)
+ *   2. (nprofile1|nevent1|naddr1)[bech32]{40,300} - variable-length TLV-encoded identifiers
+ * - [023456789acdefghjklmnpqrstuvwxyz] - bech32 character set (excludes 1, b, i, o)
+ *
+ * Length constraints prevent matching too many characters when text immediately follows:
+ * - "nostr:npub1...{58 chars}how" will match only the 63-char npub, not including "how"
+ * - This works because npub/note are always exactly 63 chars (including prefix)
+ * - Other types are TLV-encoded and vary, but typically 100-200 chars
+ */
+const NOSTR_MENTION_PATTERN =
+  /\bnostr:((?:(?:npub1|note1)[023456789acdefghjklmnpqrstuvwxyz]{58})|(?:(?:nprofile1|nevent1|naddr1)[023456789acdefghjklmnpqrstuvwxyz]{40,300}))/gi;
+
+/**
+ * Transformer that finds nostr: mentions and converts them to mention nodes.
+ * Compatible with applesauce-content's transformer pipeline.
+ *
+ * This transformer runs BEFORE the default textNoteTransformers to handle
+ * edge cases where mentions are immediately followed by text without whitespace.
+ */
+export function nostrMentionReferences() {
+  return (tree: Root) => {
+    findAndReplace(tree, [
+      [
+        NOSTR_MENTION_PATTERN,
+        (full, encoded) => {
+          try {
+            // Decode the nip19 identifier
+            const decoded = nip19.decode(encoded);
+
+            // Return a mention node with decoded data
+            return {
+              type: "mention",
+              decoded: {
+                type: decoded.type,
+                data: decoded.data,
+              },
+              encoded,
+              raw: full,
+            } as unknown as Content;
+          } catch (error) {
+            // If decode fails, return as plain text
+            console.warn(
+              `[nostr-mention-transformer] Failed to decode ${encoded}:`,
+              error,
+            );
+            return {
+              type: "text",
+              value: full,
+            } as unknown as Content;
+          }
+        },
+      ],
+    ]);
+  };
+}