diff --git a/docs/0-requirements.ja.md b/docs/0-requirements.ja.md
index e33296c..892fd3f 100644
--- a/docs/0-requirements.ja.md
+++ b/docs/0-requirements.ja.md
@@ -104,6 +104,8 @@ WebhookMcpAgent DO が以下のツールセットを提供する。ローカル
 
 **F3.1 ローカルブリッジ整形:** ローカルブリッジは `get_pending_status` の戻り値を Claude Code UserPromptSubmit hook の decision JSON shape (`hookSpecificOutput.hookEventName="UserPromptSubmit"` + `additionalContext` に pending_count / types / latest_received_at の自然文要約) にラップして返す。これは `type: "mcp_tool"` UserPromptSubmit hook 経由の呼び出しで戻り値が AI 文脈に注入されるための要件であり、手動 tool 呼び出し時も同 shape で返る。リモート (Worker + DO) 側の戻り値構造は変更しない。
 
+**F3.1 空状態の silent (empty silent, #221):** `pending_count == 0` の場合はラップせずリモート戻り値をそのまま返す。Claude Code 側で decision schema に一致しない JSON は silent discard されるため、hook 経由の呼び出しで `additionalContext` に何も注入されず、毎ターン空 reminder のノイズが消える。手動 tool 呼び出しではリモートの raw payload (`{pending_count: 0, types: {}, latest_received_at: null}`) が返り、AI は内容を直接判定できる。
+
 **イベントサマリー構造:**
 
 ```json
diff --git a/docs/0-requirements.md b/docs/0-requirements.md
index c09cd16..1d92538 100644
--- a/docs/0-requirements.md
+++ b/docs/0-requirements.md
@@ -104,6 +104,8 @@ WebhookMcpAgent DO が以下のツールセットを提供する。ローカル
 
 **F3.1 Local bridge shaping:** The local bridge wraps `get_pending_status` results into the Claude Code UserPromptSubmit hook decision JSON shape (`hookSpecificOutput.hookEventName="UserPromptSubmit"` plus a natural-language summary of pending_count / types / latest_received_at in `additionalContext`). This is required so values returned via `type: "mcp_tool"` UserPromptSubmit hooks reach the AI prompt context; manual tool calls receive the same shape. The remote (Worker + DO) return contract is unchanged.
 
+**F3.1 Empty silent (#221):** When `pending_count == 0`, the local bridge returns the remote payload untouched (no wrap). Claude Code silently discards JSON that does not match a decision schema, so hook callers receive nothing in `additionalContext` — eliminating the per-turn empty-reminder noise. Manual tool callers see the raw payload (`{pending_count: 0, types: {}, latest_received_at: null}`) and can interpret it directly.
+
 **イベントサマリー構造:**
 
 ```json
diff --git a/mcp-server/server/index.js b/mcp-server/server/index.js
index 01e0601..188705f 100644
--- a/mcp-server/server/index.js
+++ b/mcp-server/server/index.js
@@ -833,6 +833,12 @@ function formatPendingStatusSummary(payload) {
  * Manual `tool_use` callers still get a JSON object back; AI can read the
  * decision shape directly. No `decision: "block"` field is set, so the
  * user's prompt is never blocked.
+ *
+ * Empty silent (#221): when the parsed payload reports `pending_count === 0`,
+ * return the result untouched (no wrap). The raw remote JSON does not match
+ * any Claude Code decision schema, so hook callers receive nothing in
+ * `additionalContext` — eliminating the per-turn empty reminder noise. Manual
+ * callers see the raw payload, which is self-describing.
  */
 function wrapGetPendingStatusAsDecisionJson(result) {
   if (!result || !Array.isArray(result.content) || result.content.length === 0) {
@@ -844,7 +850,11 @@ function wrapGetPendingStatusAsDecisionJson(result) {
   }
   let summary;
   try {
-    summary = formatPendingStatusSummary(JSON.parse(first.text));
+    const payload = JSON.parse(first.text);
+    if (payload && typeof payload === "object" && payload.pending_count === 0) {
+      return result;
+    }
+    summary = formatPendingStatusSummary(payload);
   } catch {
     summary = first.text.slice(0, 200);
   }
diff --git a/mcp-server/test/get-pending-status-decision-shape.test.mjs b/mcp-server/test/get-pending-status-decision-shape.test.mjs
index 4c609b9..6ad57af 100644
--- a/mcp-server/test/get-pending-status-decision-shape.test.mjs
+++ b/mcp-server/test/get-pending-status-decision-shape.test.mjs
@@ -65,7 +65,11 @@ function wrapGetPendingStatusAsDecisionJson(result) {
   }
   let summary;
   try {
-    summary = formatPendingStatusSummary(JSON.parse(first.text));
+    const payload = JSON.parse(first.text);
+    if (payload && typeof payload === "object" && payload.pending_count === 0) {
+      return result;
+    }
+    summary = formatPendingStatusSummary(payload);
   } catch {
     summary = first.text.slice(0, 200);
   }
@@ -112,7 +116,7 @@ test("wraps non-empty pending payload as UserPromptSubmit decision JSON", () =>
   assert.equal("decision" in decision, false);
 });
 
-test("wraps zero-pending payload with explicit no-events sentence", () => {
+test("returns zero-pending payload untouched (empty silent, #221)", () => {
   const remote = {
     content: [
       {
@@ -126,13 +130,13 @@ test("wraps zero-pending payload with explicit no-events sentence", () => {
     ],
   };
 
-  const decision = JSON.parse(
-    wrapGetPendingStatusAsDecisionJson(remote).content[0].text,
-  );
-  assert.equal(
-    decision.hookSpecificOutput.additionalContext,
-    "No pending GitHub webhook events.",
-  );
+  const wrapped = wrapGetPendingStatusAsDecisionJson(remote);
+  assert.deepEqual(wrapped, remote);
+  // No decision schema is produced; hook callers receive nothing in
+  // additionalContext, eliminating the per-turn empty reminder noise.
+  const parsed = JSON.parse(wrapped.content[0].text);
+  assert.equal(parsed.pending_count, 0);
+  assert.equal("hookSpecificOutput" in parsed, false);
 });
 
 test("falls back to truncated raw text when remote payload is not JSON", () => {
@@ -169,7 +173,7 @@ test("preserves sibling result fields like isError when wrapping", () => {
     content: [
       {
         type: "text",
-        text: JSON.stringify({ pending_count: 0, types: {} }),
+        text: JSON.stringify({ pending_count: 3, types: { issues: 3 } }),
       },
     ],
   };