diff --git a/src/crypto_strategies/entrypoints/__init__.py b/src/crypto_strategies/entrypoints/__init__.py
index f2e9af8..91b2d9d 100644
--- a/src/crypto_strategies/entrypoints/__init__.py
+++ b/src/crypto_strategies/entrypoints/__init__.py
@@ -18,6 +18,8 @@
     crypto_trend_rotation_manifest,
 )
 
+from ._common import apply_risk_gate
+
 
 """Unified crypto strategy entrypoints built on top of legacy core/rotation modules."""
 
@@ -253,12 +255,13 @@ def evaluate_crypto_live_pool_rotation(ctx: StrategyContext) -> StrategyDecision
             **dict(ctx.artifacts.get("trend_pool_contract", {})),
         },
     }
-    return StrategyDecision(
+    decision = StrategyDecision(
         positions=tuple(positions),
         budgets=budget_intents,
         risk_flags=risk_flags,
         diagnostics=diagnostics,
     )
+    return apply_risk_gate(decision)
 
 
 crypto_live_pool_rotation_entrypoint = CallableStrategyEntrypoint(
@@ -334,12 +337,13 @@ def evaluate_crypto_btc_dca(ctx: StrategyContext) -> StrategyDecision:
         "mayer_multiple": metadata.get("mayer_multiple", float("nan")),
     }
 
-    return StrategyDecision(
+    decision = StrategyDecision(
         positions=tuple(positions),
         budgets=budget_intents,
         risk_flags=risk_flags,
         diagnostics=diagnostics,
     )
+    return apply_risk_gate(decision, max_single_weight=0.50)
 
 
 crypto_btc_dca_entrypoint = CallableStrategyEntrypoint(
@@ -416,12 +420,13 @@ def evaluate_crypto_trend_rotation(ctx: StrategyContext) -> StrategyDecision:
         "profile": metadata.get("profile"),
     }
 
-    return StrategyDecision(
+    decision = StrategyDecision(
         positions=tuple(positions),
         budgets=budget_intents,
         risk_flags=risk_flags,
         diagnostics=diagnostics,
     )
+    return apply_risk_gate(decision, max_single_weight=0.50)
 
 
 crypto_trend_rotation_entrypoint = CallableStrategyEntrypoint(
@@ -604,12 +609,13 @@ def evaluate_crypto_equity_combo(ctx: StrategyContext) -> StrategyDecision:
         },
     }
 
-    return StrategyDecision(
+    decision = StrategyDecision(
         positions=tuple(positions),
         budgets=budget_intents,
         risk_flags=risk_flags,
         diagnostics=diagnostics,
     )
+    return apply_risk_gate(decision)
 
 
 crypto_equity_combo_entrypoint = CallableStrategyEntrypoint(
diff --git a/src/crypto_strategies/entrypoints/_common.py b/src/crypto_strategies/entrypoints/_common.py
new file mode 100644
index 0000000..4101a85
--- /dev/null
+++ b/src/crypto_strategies/entrypoints/_common.py
@@ -0,0 +1,115 @@
+from __future__ import annotations
+
+import logging
+
+from quant_platform_kit.strategy_contracts import PositionTarget, StrategyDecision
+
+logger = logging.getLogger(__name__)
+
+# ---------------------------------------------------------------------------
+# 风控硬门 — 每个 entrypoint 返回 StrategyDecision 前必须调用
+# ---------------------------------------------------------------------------
+
+
+def _position_weight(position: PositionTarget) -> float | None:
+    if position.target_weight is not None:
+        return abs(float(position.target_weight))
+    return None
+
+
+def _reject_risk_gate(
+    decision: StrategyDecision,
+    *,
+    risk_flag: str,
+    reason: str,
+) -> StrategyDecision:
+    return StrategyDecision(
+        positions=(),
+        budgets=decision.budgets,
+        risk_flags=(*decision.risk_flags, risk_flag),
+        diagnostics={
+            **(decision.diagnostics or {}),
+            "risk_gate": "REJECT",
+            "reason": reason,
+        },
+    )
+
+
+def apply_risk_gate(
+    decision: StrategyDecision,
+    *,
+    max_single_weight: float = 1.0,
+    max_positions: int = 20,
+    max_total_exposure: float = 1.0,
+) -> StrategyDecision:
+    """对所有 StrategyDecision 施加硬风控门。
+
+    检查项：
+    1. 单仓位集中度（> max_single_weight → REJECT，默认 100% 即不限制）
+    2. 持仓数量（> max_positions → REJECT）
+    3. 总仓位超限（> max_total_exposure → REJECT）
+
+    各策略类型可根据自身特点调整门限：
+    - ETF 轮动：max_single_weight=1.0（ETF 本身就是分散的篮子）
+    - 个股精选：max_single_weight=0.10
+    - 加密货币：max_single_weight=0.20, max_positions=10
+
+    如果 REJECT，返回空仓决策并标注拒绝原因。
+    这个函数不可绕过 —— AGENTS.md 要求所有 entrypoint 必须调用。
+    """
+    positions = decision.positions or ()
+
+    if not positions:
+        return StrategyDecision(
+            positions=decision.positions,
+            budgets=decision.budgets,
+            risk_flags=decision.risk_flags,
+            diagnostics={**(decision.diagnostics or {}), "risk_gate": "APPROVE"},
+        )
+
+    weight_positions = [p for p in positions if _position_weight(p) is not None]
+
+    if max_single_weight < 1.0 and weight_positions:
+        for p in weight_positions:
+            weight = _position_weight(p)
+            assert weight is not None
+            if weight > max_single_weight:
+                logger.warning(
+                    "risk_gate REJECT concentration: symbol=%s weight=%.2f%% limit=%.0f%%",
+                    p.symbol, weight * 100, max_single_weight * 100,
+                )
+                return _reject_risk_gate(
+                    decision,
+                    risk_flag="rejected:concentration",
+                    reason=f"{p.symbol} {weight:.1%} > {max_single_weight:.0%} 上限",
+                )
+
+    if len(positions) > max_positions:
+        logger.warning(
+            "risk_gate REJECT position_count: %d > %d", len(positions), max_positions,
+        )
+        return _reject_risk_gate(
+            decision,
+            risk_flag="rejected:too_many_positions",
+            reason=f"{len(positions)} 个持仓 > {max_positions} 上限",
+        )
+
+    if weight_positions:
+        total_weight = sum(_position_weight(p) or 0.0 for p in weight_positions)
+        if total_weight > max_total_exposure + 1e-9:
+            logger.warning(
+                "risk_gate REJECT total_exposure: %.2f%% > %.0f%%",
+                total_weight * 100, max_total_exposure * 100,
+            )
+            return _reject_risk_gate(
+                decision,
+                risk_flag="rejected:overexposed",
+                reason=f"总仓位 {total_weight:.1%} > {max_total_exposure:.0%}",
+            )
+
+    return StrategyDecision(
+        positions=decision.positions,
+        budgets=decision.budgets,
+        risk_flags=decision.risk_flags,
+        diagnostics={**(decision.diagnostics or {}), "risk_gate": "APPROVE"},
+    )