feat: skill triger bug

[Wang-Daoji]

Description

Please include a summary of the change, the problem it solves, the implementation approach, and relevant context. List any dependencies required for this change.

Related Issue (Required): Fixes #1769

Type of change

Please delete options that are not relevant.

• Bug fix (non-breaking change which fixes an issue)

How Has This Been Tested?

import time
import json
from memos.api.routers.server_router import add_memories, search_memories
from memos.api.product_models import APIADDRequest, APISearchRequest


messages = TOOL_CALL_SKILL_EXAMPLES_FIVE_ROUNDS

if __name__ == "__main__":
    user_id = f"test_user_{time.time()}"
    # user_id = "test_user_456765aax43156789gg921336533nn08khb21122"
    # user_id = 'test_user_1770630514.518709'
    readable_cube_ids = [user_id]
    add_req = APIADDRequest(
        user_id=user_id,
        writable_cube_ids=readable_cube_ids,
        messages=[
    {
        "role": "system",
        "content": "你是一个生产故障排查助手。需要使用工具获取日志、指标、链路、配置和历史工单后，再输出结构化排查结论。",
    },
    {
        "role": "user",
        "content": "推荐服务 recommendation-api 今天 10 点左右延迟突然升高，帮我排查根因，并总结以后同类问题应该怎么查。",
    },
    {
        "role": "assistant",
        "tool_calls": [
            {
                "id": "call_metrics_001",
                "type": "function",
                "function": {
                    "name": "query_metrics",
                    "arguments": '{"service":"recommendation-api","time_window":"2026-04-18 09:40-10:20 CST","metrics":["p95_latency","p99_latency","error_rate","request_count"]}',
                },
            }
        ],
    },
    {
        "role": "tool",
        "tool_call_id": "call_metrics_001",
        "content": '{"service":"recommendation-api","time_window":"2026-04-18 09:40-10:20 CST","p95_latency":"320ms -> 2.8s","p99_latency":"610ms -> 6.4s","error_rate":"0.2% -> 8.7%","request_count":"stable at 4200 rpm","first_anomaly_at":"2026-04-18 10:03 CST"}',
    },
    {
        "role": "assistant",
        "tool_calls": [
            {
                "id": "call_logs_002",
                "type": "function",
                "function": {
                    "name": "search_logs",
                    "arguments": '{"service":"recommendation-api","time_window":"2026-04-18 10:00-10:12 CST","levels":["ERROR","WARN"],"keywords":["timeout","redis","feature_flag"]}',
                },
            }
        ],
    },
    {
        "role": "tool",
        "tool_call_id": "call_logs_002",
        "content": '{"matches":[{"time":"2026-04-18 10:03:11 CST","level":"WARN","message":"RedisTimeout: feature_cache get timeout after 800ms"},{"time":"2026-04-18 10:03:13 CST","level":"ERROR","message":"fallback model scoring failed because feature vector is incomplete"},{"time":"2026-04-18 10:04:01 CST","level":"WARN","message":"circuit breaker open for feature_cache"}],"summary":"错误集中在 feature_cache 读取超时和降级链路失败。"}',
    },
    {
        "role": "assistant",
        "tool_calls": [
            {
                "id": "call_trace_003",
                "type": "function",
                "function": {
                    "name": "query_traces",
                    "arguments": '{"service":"recommendation-api","time_window":"2026-04-18 10:00-10:12 CST","sample_type":"slow_and_error","limit":20}',
                },
            }
        ],
    },
    {
        "role": "tool",
        "tool_call_id": "call_trace_003",
        "content": '{"slow_spans":[{"span":"GET /recommend","p95":"2.9s","child_span":"feature-service.batchGet","child_p95":"2.4s"},{"span":"feature-service.batchGet","dependency":"redis://feature-cache","p95":"2.1s"}],"error_traces":"18/20 traces fail after feature_cache timeout, then fallback scoring receives empty feature vector."}',
    },
    {
        "role": "assistant",
        "tool_calls": [
            {
                "id": "call_deploy_004",
                "type": "function",
                "function": {
                    "name": "get_deploy_diff",
                    "arguments": '{"service":"recommendation-api","around_time":"2026-04-18 10:03 CST","include":["deployments","config_changes","feature_flags"]}',
                },
            }
        ],
    },
    {
        "role": "tool",
        "tool_call_id": "call_deploy_004",
        "content": '{"deployments":[],"config_changes":[{"time":"2026-04-18 09:58 CST","owner":"search-platform","key":"feature_cache.timeout_ms","old_value":"300","new_value":"800"},{"time":"2026-04-18 09:59 CST","owner":"ml-platform","key":"fallback_scoring.required_features","old_value":"allow_missing","new_value":"strict"}],"feature_flags":[{"time":"2026-04-18 10:00 CST","flag":"new_feature_cache_key_v2","status":"enabled","scope":"100% traffic"}]}',
    },
    # {
    #     "role": "assistant",
    #     "tool_calls": [
    #         {
    #             "id": "call_incidents_005",
    #             "type": "function",
    #             "function": {
    #                 "name": "search_incident_history",
    #                 "arguments": '{"keywords":["feature_cache","new_feature_cache_key_v2","fallback_scoring","redis timeout"],"lookback_days":90}',
    #             },
    #         }
    #     ],
    # },
    # {
    #     "role": "tool",
    #     "tool_call_id": "call_incidents_005",
    #     "content": '{"related_incidents":[{"date":"2026-03-22","service":"ranking-api","cause":"new cache key format caused 35% miss rate; fallback tolerated missing features so impact was hidden","fix":"rolled back key format flag, warmed cache, added miss-rate alert"},{"date":"2026-02-09","service":"recommendation-api","cause":"fallback scoring strict mode exposed incomplete feature vector","fix":"changed fallback policy to allow missing non-critical features"}],"suggested_checks":["compare cache hit rate by key version","rollback new_feature_cache_key_v2","temporarily set fallback_scoring.required_features=allow_missing"]}',
    # },
    {
        "role": "assistant",
        "content": "# recommendation-api 延迟升高排查结论\n\n## 现象\n- 2026-04-18 10:03 CST 开始，p95 从 320ms 升至 2.8s，p99 升至 6.4s，错误率升至 8.7%。\n- 请求量稳定，说明不是流量突增导致。\n\n## 根因判断\n1. 直接瓶颈在 `feature_cache`：日志和链路都显示 `feature-service.batchGet` 访问 Redis feature cache 超时。\n2. 触发因素是 10:00 全量打开 `new_feature_cache_key_v2`，很可能造成缓存 key 变更后的命中率下降。\n3. 影响被放大是因为 09:59 同时把 `fallback_scoring.required_features` 从 `allow_missing` 改成 `strict`，导致缓存超时后降级模型无法接受缺失特征，错误率随延迟一起上升。\n\n## 建议动作\n1. 立即回滚 `new_feature_cache_key_v2` 到关闭状态，观察 p95、错误率、feature cache 命中率是否恢复。\n2. 临时把 `fallback_scoring.required_features` 改回 `allow_missing`，先恢复服务可用性。\n3. 回滚后补充验证：按 key 版本对比 cache hit rate、Redis timeout、fallback scoring error 三个指标。\n4. 复盘时补一条发布护栏：缓存 key 变更必须先灰度并预热缓存，且不能和降级策略收紧同时发布。\n\n## 可复用排障流程\n以后排查类似线上问题，先锁定异常时间窗和用户影响，再用指标判断是流量、依赖还是错误率问题；随后用日志确认错误类型，用 trace 定位慢 span 和依赖，再查部署、配置、feature flag 的时间线，最后用历史工单验证是否有相同模式。输出结论时要分清直接瓶颈、触发因素、放大因素，并给出回滚、验证和长期防护动作。",
    },
    {
        "role": "user",
        "content": "以后这类生产故障分析，请固定输出“现象、根因判断、立即止血、验证指标、长期防护”五个模块，并且要明确区分触发因素和放大因素。",
    },
    {
        "role": "assistant",
        "content": "收到。以后处理生产故障排查类任务，我会先通过指标、日志、链路、变更记录和历史工单形成证据链，再固定按“现象、根因判断、立即止血、验证指标、长期防护”五个模块输出；在根因判断里会明确拆分直接瓶颈、触发因素和放大因素，避免把相关现象混成单一原因。",
    },
]
,
        async_mode="sync",
        mode="fine",
        info={"sbbbbb": "asdsdadsad", "bhjjjjjjjjjjj": "uuueueueueueueue"},
        session_id="1234567890",
        manager_user_id="super_manager_xxx",
        project_id="super_project_xxx",
    )

    add_rsp = add_memories(add_req)
    print("add_memories rsp: \n\n", add_rsp)

    time.sleep(5)

    search_req = APISearchRequest(
        user_id=user_id,
        readable_cube_ids=readable_cube_ids,
        query="线上服务延迟升高和错误率上升时如何排查根因",
        include_preference=True,
        search_tool_memory=True,
        include_skill_memory=True,
    )
    start = time.time()
    search_rsp = search_memories(search_req).data["skill_mem"]
    end = time.time()
    print(f"search_memories cost time: {end - start}")

    print("\n\nsearch_rsp: \n\n", json.dumps(search_rsp, indent=2, ensure_ascii=False))

Please describe the tests that you ran to verify your changes. Provide instructions so we can reproduce. Please also list any relevant details for your test configuration

• Unit Test
• Test Script Or Test Steps (please provide)
• Pipeline Automated API Test (please provide)

Checklist

• I have performed a self-review of my own code | 我已自行检查了自己的代码
• I have commented my code in hard-to-understand areas | 我已在难以理解的地方对代码进行了注释
• I have added tests that prove my fix is effective or that my feature works | 我已添加测试以证明我的修复有效或功能正常
• I have created related documentation issue/PR in MemOS-Docs (if applicable) | 我已在 MemOS-Docs 中创建了相关的文档 issue/PR（如果适用）
• I have linked the issue to this PR (if applicable) | 我已将 issue 链接到此 PR（如果适用）
• I have mentioned the person who will review this PR | 我已提及将审查此 PR 的人

Reviewer Checklist

• closes #xxxx (Replace xxxx with the GitHub issue number)
• Made sure Checks passed
• Tests have been provided