diff --git a/wg1/gen-worker/gamedef_quickcheck.py b/wg1/gen-worker/gamedef_quickcheck.py
index f63fa1a2..6007c0e6 100644
--- a/wg1/gen-worker/gamedef_quickcheck.py
+++ b/wg1/gen-worker/gamedef_quickcheck.py
@@ -27,6 +27,7 @@ SAA_PROMPTS = (REPO_ROOT / "game-cloud/game-module-aigc/game-module-aigc-server/
 GAME_RUNTIME = REPO_ROOT / "game-runtime"
 GEN_DIR = GAME_RUNTIME / "games" / "_wg1-gen"
 BRIEFS = Path(__file__).resolve().parent / "briefs"
+MAX_RETRIES = 2   # repair 回环上限（mirror studio.py max_repairs 默认；总尝试=MAX_RETRIES+1）
 
 
 def extract_block(name):
@@ -102,33 +103,39 @@ def run_one(game_id, gamedef_system, design_system, model):
             out["driver"] = (gs.get("driver") or {}).get("type")
     except Exception as e:
         out["err"] = "design 降级:" + str(e)[:80]   # design 失败不致命，凭 brief 继续 generate
-    # ② generate gamedef（用 enriched=brief+设计稿）
-    user = "请为下面这款游戏产出一份 gameDefinition（只输出一个 JSON 对象，不要 ```代码块、不要解释）：\n\n" + enriched
-    try:
-        resp = _client.chat(model, gamedef_system, user, max_tokens=16000)
-    except Exception as e:
-        out["stage"] = "generate"; out["err"] = str(e)[:200]; return out
-    out["tokens"] = (resp.get("prompt_tokens"), resp.get("completion_tokens"))
-    gd = extract_json_obj(resp["content"])
-    if gd is None:
-        out["stage"] = "parse"; out["err"] = "未解析出 gameDefinition JSON"; return out
-    ok, err = assemble_via_cli(gid, gd)
-    if not ok:
-        out["stage"] = "validate/assemble"; out["err"] = err[:400]; return out
-    # 复用 run.py 的 scaffold 余下（拷模板 + play-spec）+ build + play（factory 已由 CLI 写好）。
+    # ② generate→assemble→build→play **repair 回环**（mirror 真图：失败→verdict/错误回喂→重生成,≤max_retries 次）。
+    #    这是 ≥60% cutover 门的真机制（单发首攻 ≠ 终率）；design 只跑一次,repair 只重 generate。
     gdir = GEN_DIR / gid
-    shutil.copyfile(GEN_DIR / "_shared" / "entry-bundle.template.js", gdir / "entry-bundle.js")
-    shutil.copyfile(GEN_DIR / "_shared" / "index.template.html", gdir / "index.html")
-    (gdir / "play-spec.json").write_text(json.dumps(play_spec, ensure_ascii=False, indent=2), encoding="utf-8")
-    bok, blog = run.build(gid)
-    if not bok:
-        out["stage"] = "build"; out["err"] = blog[:400]; return out
-    rc, _, verdict = run.play(gid)
-    out["stage"] = "play"
-    out["guards"] = {k: g.get("pass") for k, g in ((verdict or {}).get("guards") or {}).items()}
-    out["pass"] = bool(rc == 0 and verdict and verdict.get("pass"))
-    if not out["pass"]:
-        out["err"] = run._verdict_feedback(verdict)[:300]
+    feedback = None
+    for attempt in range(MAX_RETRIES + 1):
+        out["attempts"] = attempt + 1
+        user = "请为下面这款游戏产出一份 gameDefinition（只输出一个 JSON 对象，不要 ```代码块、不要解释）：\n\n" + enriched
+        if feedback:
+            user += "\n\n———\n上一次生成【未通过】，原因如下，请针对性修正后重新产出完整 gameDefinition JSON：\n" + feedback
+        try:
+            resp = _client.chat(model, gamedef_system, user, max_tokens=16000)
+        except Exception as e:
+            out["stage"] = "generate"; feedback = "模型调用失败：" + str(e)[:160]; continue
+        gd = extract_json_obj(resp["content"])
+        if gd is None:
+            out["stage"] = "parse"; feedback = "未能解析出 gameDefinition JSON。请只输出一个 JSON 对象（顶层 entities/components/behaviors/scenes/rules）。"; continue
+        ok, err = assemble_via_cli(gid, gd)
+        if not ok:
+            out["stage"] = "validate/assemble"; feedback = "校验/装配未过：\n" + err[:500]; continue
+        shutil.copyfile(GEN_DIR / "_shared" / "entry-bundle.template.js", gdir / "entry-bundle.js")
+        shutil.copyfile(GEN_DIR / "_shared" / "index.template.html", gdir / "index.html")
+        (gdir / "play-spec.json").write_text(json.dumps(play_spec, ensure_ascii=False, indent=2), encoding="utf-8")
+        bok, blog = run.build(gid)
+        if not bok:
+            out["stage"] = "build"; feedback = "打包失败（esbuild）：\n" + blog[:500]; continue
+        rc, _, verdict = run.play(gid)
+        out["stage"] = "play"
+        out["guards"] = {k: g.get("pass") for k, g in ((verdict or {}).get("guards") or {}).items()}
+        if rc == 0 and verdict and verdict.get("pass"):
+            out["pass"] = True
+            break
+        feedback = run._verdict_feedback(verdict)
+        out["err"] = feedback[:300]
     return out
 
 
@@ -146,7 +153,8 @@ def main():
         gpass = sum(1 for v in r["guards"].values() if v)
         gtot = len(r["guards"])
         drv = r.get("driver") or "-"
-        print(f"  {r['game_id']:14s} {mark:16s} drv={drv:16s} gates={gpass}/{gtot}  {('' if r['pass'] else r['err'][:110])}")
+        att = r.get("attempts", 1)
+        print(f"  {r['game_id']:14s} {mark:16s} drv={drv:14s} att={att} gates={gpass}/{gtot}  {('' if r['pass'] else r['err'][:100])}")
     npass = sum(1 for r in results if r["pass"])
     print(f"\n[gamedef-qc] 过门 {npass}/{len(results)}  (cutover 门基线=iife 60%; 本为 lili-mac 小样早读, authoritative=mini-desktop)")
     Path(__file__).resolve().parent.joinpath("results", "gamedef-quickcheck.json").write_text(