fix(agents): strip markdown code spans from IDENTITY.md values and labels

Refs #86594

src/agents/identity-file.test.ts

@@ -33,6 +33,26 @@ describe("parseIdentityMarkdown", () => {
       avatar: "avatars/openclaw.png",
     });
   });
+
+  it("strips markdown code spans from values and labels", () => {
+    const content = [
+      "- **Name:** `Samantha`",
+      "- `Creature`: Robot",
+      "- **`Avatar`**: `avatars/openclaw.png`",
+    ].join("\n");
+    const parsed = parseIdentityMarkdown(content);
+    expect(parsed).toEqual({
+      name: "Samantha",
+      creature: "Robot",
+      avatar: "avatars/openclaw.png",
+    });
+  });
+
+  it("still treats code-span-wrapped template placeholders as placeholders", () => {
+    const content = "- **Avatar:** `(workspace-relative path, http(s) URL, or data URI)`";
+    const parsed = parseIdentityMarkdown(content);
+    expect(parsed).toStrictEqual({});
+  });
 });
 
 describe("mergeIdentityMarkdownContent", () => {

src/agents/identity-file.ts

@@ -31,7 +31,7 @@ const IDENTITY_PLACEHOLDER_VALUES = new Set([
 
 function normalizeIdentityValue(value: string): string {
   let normalized = value.trim();
-  normalized = normalized.replace(/^[*_]+|[*_]+$/g, "").trim();
+  normalized = normalized.replace(/^[*_`\s]+|[*_`\s]+$/g, "").trim();
   if (normalized.startsWith("(") && normalized.endsWith(")")) {
     normalized = normalized.slice(1, -1).trim();
   }
@@ -54,11 +54,11 @@ export function parseIdentityMarkdown(content: string): AgentIdentityFile {
       continue;
     }
     const label = normalizeLowercaseStringOrEmpty(
-      cleaned.slice(0, colonIndex).replace(/[*_]/g, ""),
+      cleaned.slice(0, colonIndex).replace(/[*_`]/g, ""),
     );
     const value = cleaned
       .slice(colonIndex + 1)
-      .replace(/^[*_]+|[*_]+$/g, "")
+      .replace(/^[*_`\s]+|[*_`\s]+$/g, "")
       .trim();
     if (!value) {
       continue;