diff --git a/CHANGELOG.md b/CHANGELOG.md
index 8d8aeaf02..5f50da6cd 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -35,6 +35,7 @@ Docs: https://docs.openclaw.ai
 - Agents/Overflow: add Chinese context-overflow pattern detection in `isContextOverflowError` so localized provider errors route through overflow recovery paths. (#22855) Thanks @Clawborn.
 - Agents/Failover: treat HTTP 502/503/504 errors as failover-eligible transient timeouts so fallback chains can switch providers/models during upstream outages instead of retrying the same failing target. (#20999) Thanks @taw0002 and @vincentkoc.
 - Auto-reply/Inbound metadata: hide direct-chat `message_id`/`message_id_full` and sender metadata only from normalized chat type (not sender-id sentinels), preserving group metadata visibility and preventing sender-id spoofed direct-mode classification. (#24373) thanks @jd316.
+- Auto-reply/Inbound metadata: move dynamic inbound `flags` (reply/forward/thread/history) from system metadata to user-context conversation info, preventing turn-by-turn prompt-cache invalidation from flag toggles. (#21785) Thanks @aidiffuser.
 - Auto-reply/Sessions: remove auth-key labels from `/new` and `/reset` confirmation messages so session reset notices never expose API key prefixes or env-key labels in chat output. (#24384, #24409) Thanks @Clawborn.
 - Slack/Group policy: move Slack account `groupPolicy` defaulting to provider-level schema defaults so multi-account configs inherit top-level `channels.slack.groupPolicy` instead of silently overriding inheritance with per-account `allowlist`. (#17579) Thanks @ZetiMente.
 - Providers/Anthropic: skip `context-1m-*` beta injection for OAuth/subscription tokens (`sk-ant-oat-*`) while preserving OAuth-required betas, avoiding Anthropic 401 auth failures when `params.context1m` is enabled. (#10647, #20354) Thanks @ClumsyWizardHands and @dcruver.
diff --git a/src/auto-reply/reply/inbound-meta.test.ts b/src/auto-reply/reply/inbound-meta.test.ts
index 239aa23bc..a85cbadab 100644
--- a/src/auto-reply/reply/inbound-meta.test.ts
+++ b/src/auto-reply/reply/inbound-meta.test.ts
@@ -57,6 +57,24 @@ describe("buildInboundMetaSystemPrompt", () => {
     expect(payload["sender_id"]).toBeUndefined();
   });
 
+  it("does not include per-turn flags in system metadata", () => {
+    const prompt = buildInboundMetaSystemPrompt({
+      ReplyToBody: "quoted",
+      ForwardedFrom: "sender",
+      ThreadStarterBody: "starter",
+      InboundHistory: [{ sender: "a", body: "b", timestamp: 1 }],
+      WasMentioned: true,
+      OriginatingTo: "telegram:-1001249586642",
+      OriginatingChannel: "telegram",
+      Provider: "telegram",
+      Surface: "telegram",
+      ChatType: "group",
+    } as TemplateContext);
+
+    const payload = parseInboundMetaPayload(prompt);
+    expect(payload["flags"]).toBeUndefined();
+  });
+
   it("omits sender_id when blank", () => {
     const prompt = buildInboundMetaSystemPrompt({
       MessageSid: "458",
@@ -183,6 +201,25 @@ describe("buildInboundUserContextPrefix", () => {
     expect(conversationInfo["sender_id"]).toBe("289522496");
   });
 
+  it("includes dynamic per-turn flags in conversation info", () => {
+    const text = buildInboundUserContextPrefix({
+      ChatType: "group",
+      WasMentioned: true,
+      ReplyToBody: "quoted",
+      ForwardedFrom: "sender",
+      ThreadStarterBody: "starter",
+      InboundHistory: [{ sender: "a", body: "b", timestamp: 1 }],
+    } as TemplateContext);
+
+    const conversationInfo = parseConversationInfoPayload(text);
+    expect(conversationInfo["is_group_chat"]).toBe(true);
+    expect(conversationInfo["was_mentioned"]).toBe(true);
+    expect(conversationInfo["has_reply_context"]).toBe(true);
+    expect(conversationInfo["has_forwarded_context"]).toBe(true);
+    expect(conversationInfo["has_thread_starter"]).toBe(true);
+    expect(conversationInfo["history_count"]).toBe(1);
+  });
+
   it("trims sender_id in conversation info", () => {
     const text = buildInboundUserContextPrefix({
       ChatType: "group",
diff --git a/src/auto-reply/reply/inbound-meta.ts b/src/auto-reply/reply/inbound-meta.ts
index 80a2d3c3c..418a42859 100644
--- a/src/auto-reply/reply/inbound-meta.ts
+++ b/src/auto-reply/reply/inbound-meta.ts
@@ -16,9 +16,9 @@ export function buildInboundMetaSystemPrompt(ctx: TemplateContext): string {
 
   // Keep system metadata strictly free of attacker-controlled strings (sender names, group subjects, etc.).
   // Those belong in the user-role "untrusted context" blocks.
-  // Per-message identifiers (message_id, reply_to_id, sender_id) are also excluded here: they change
-  // on every turn and would bust prefix-based prompt caches on local model providers. They are
-  // included in the user-role conversation info block via buildInboundUserContextPrefix() instead.
+  // Per-message identifiers and dynamic flags are also excluded here: they change on turns/replies
+  // and would bust prefix-based prompt caches on providers that use stable system prefixes.
+  // They are included in the user-role conversation info block instead.
 
   // Resolve channel identity: prefer explicit channel, then surface, then provider.
   // For webchat/Hub Chat sessions (when Surface is 'webchat' or undefined with no real channel),
@@ -43,14 +43,6 @@ export function buildInboundMetaSystemPrompt(ctx: TemplateContext): string {
     provider: safeTrim(ctx.Provider),
     surface: safeTrim(ctx.Surface),
     chat_type: chatType ?? (isDirect ? "direct" : undefined),
-    flags: {
-      is_group_chat: !isDirect ? true : undefined,
-      was_mentioned: ctx.WasMentioned === true ? true : undefined,
-      has_reply_context: Boolean(ctx.ReplyToBody),
-      has_forwarded_context: Boolean(ctx.ForwardedFrom),
-      has_thread_starter: Boolean(safeTrim(ctx.ThreadStarterBody)),
-      history_count: Array.isArray(ctx.InboundHistory) ? ctx.InboundHistory.length : 0,
-    },
   };
 
   // Keep the instructions local to the payload so the meaning survives prompt overrides.
@@ -92,7 +84,15 @@ export function buildInboundUserContextPrefix(ctx: TemplateContext): string {
     group_space: safeTrim(ctx.GroupSpace),
     thread_label: safeTrim(ctx.ThreadLabel),
     is_forum: ctx.IsForum === true ? true : undefined,
+    is_group_chat: !isDirect ? true : undefined,
     was_mentioned: ctx.WasMentioned === true ? true : undefined,
+    has_reply_context: ctx.ReplyToBody ? true : undefined,
+    has_forwarded_context: ctx.ForwardedFrom ? true : undefined,
+    has_thread_starter: safeTrim(ctx.ThreadStarterBody) ? true : undefined,
+    history_count:
+      Array.isArray(ctx.InboundHistory) && ctx.InboundHistory.length > 0
+        ? ctx.InboundHistory.length
+        : undefined,
   };
   if (Object.values(conversationInfo).some((v) => v !== undefined)) {
     blocks.push(