@clickzetta/cz-cli-darwin-x64 0.3.94 → 0.3.96

package/bin/skills/clickzetta-query-optimizer/references/hints-and-sortkey.md

@@ -1,78 +0,0 @@
-# Map Join 与 Sort Key 推荐参考
-
-> 来源：https://www.yunqi.tech/documents/mapjoin 和 https://www.yunqi.tech/documents/auto-index
-
----
-
-## Map Join（小表广播优化）
-
-### 语法
-
-```sql
-SELECT /*+ MAPJOIN (small_table_alias) */ *
-FROM large_table t1
-JOIN small_table t2 ON t1.id = t2.id;
-```
-
-### 说明
-
-- 将小表广播到各节点，在 Map 阶段完成 JOIN，避免 Shuffle
-- **小表大小限制：1GB**，超过则 Map Join 失败或退化为普通 JOIN
-- 适用于小表 JOIN 大表，不适用于大表 JOIN 大表
-
-### 示例
-
-```sql
--- 员工与部门关联
-SELECT /*+ MAPJOIN (dept) */ *
-FROM employees emp
-JOIN departments dept ON emp.dept_id = dept.dept_id;
-
--- 订单与客户关联
-SELECT /*+ MAPJOIN (customer) */ *
-FROM orders o
-JOIN customers customer ON o.customer_id = customer.customer_id;
-```
-
----
-
-## Sort Key 推荐（自动索引建议）
-
-### 启用自动收集
-
-```sql
--- 按天收集（推荐）
-ALTER WORKSPACE workspace_name SET PROPERTIES (auto_index='day');
-
--- 自定义参数：天/月, 最近N分钟job, 最少重复次数, 最多job数
-ALTER WORKSPACE workspace_name SET PROPERTIES (auto_index='day,150,5,100');
-```
-
-参数说明：
-- 第 1 个参数：`day`（每天）或 `month`（每月 1 号），收集时间为晚上 6 点
-- 第 2 个参数：使用最近多少分钟的 job（默认 150）
-- 第 3 个参数：job 需要重复多少次才被采用（默认 5）
-- 第 4 个参数：每列最多使用的 job 数（默认 100）
-
-### 查询推荐结果
-
-```sql
-SELECT * FROM information_schema.sortkey_candidates;
-```
-
-返回字段：`table_name`、`col`（推荐列）、`statement`（可直接执行的 ALTER 语句）、`ratio`（估算提升效果百分比）
-
-### 应用推荐
-
-```sql
--- 直接执行 statement 列中的 SQL 即可设置 sort key
-ALTER TABLE schema.table_name SET PROPERTIES("hint.sort.columns"="column_name");
-```
-
-### 建议
-
-执行前先对表收集统计信息，提高推荐准确性：
-
-```sql
-ANALYZE TABLE schema.table_name;
-```

package/bin/skills/clickzetta-query-optimizer/references/optimize.md

@@ -1,65 +0,0 @@
-# OPTIMIZE 命令参考
-
-> 来源：https://www.yunqi.tech/documents/OPTIMIZE 和 https://www.yunqi.tech/documents/small_file_optimization
-
-## 语法
-
-```sql
-OPTIMIZE table_name
-[WHERE predicate]
-[OPTIONS('key' = 'value')]
-```
-
-## 参数说明
-
-- `table_name`：格式为 `[schema_name.]table_name`
-- `WHERE predicate`：（可选）分区过滤条件，必须包含完整分区列匹配
-  - 格式：`partition_column = 'value'` 或 `dt='2023-01-01' AND region='us'`
-- `OPTIONS`：（可选）控制执行模式
-
-## 执行模式
-
-### 异步模式（默认）
-
-立即返回 Job ID，后台执行，不阻塞当前连接。
-
-```sql
--- 默认异步
-OPTIMIZE my_schema.orders;
-
--- 显式指定异步
-OPTIMIZE my_schema.orders OPTIONS('cz.sql.optimize.table.async' = 'true');
-```
-
-### 同步模式
-
-阻塞直到完成，适合开发测试和小表优化。
-
-```sql
-OPTIMIZE my_schema.orders OPTIONS('cz.sql.optimize.table.async' = 'false');
-```
-
-## 核心功能
-
-- **小文件合并**：将多个小文件整合为大文件，减少文件元数据开销
-- **删除标记清理**：清理 UPDATE/DELETE 产生的删除标记，回收存储空间
-- **数据重组**：重新整理数据布局，提升查询性能
-
-## 注意事项
-
-- **只能在通用型计算集群（GENERAL PURPOSE VIRTUAL CLUSTER）运行**，分析型集群不生效
-- 后台默认会不定时自动执行文件合并，手动 OPTIMIZE 用于精细控制
-
-## DML 写入时自动触发小文件合并
-
-```sql
--- 在 DML 执行时同时触发小文件合并
-SET cz.sql.compaction.after.commit = true;
-INSERT INTO my_table VALUES (...);
-```
-
-## 查看分区文件数量
-
-```sql
-SHOW PARTITIONS EXTENDED table_name;
-```

package/bin/skills/clickzetta-query-optimizer/references/result-cache.md

@@ -1,49 +0,0 @@
-# Result Cache（查询结果缓存）参考
-
-> 来源：https://www.yunqi.tech/documents/result_cache
-
-## 概述
-
-ClickZetta Lakehouse 提供三种缓存：
-1. **查询结果缓存（Result Cache）** — 本文档
-2. 元数据缓存（Metadata Cache）— 工作空间内共享
-3. 虚拟集群本地缓存（Local Disk Cache）— 仅限指定集群
-
-## 启用 / 禁用
-
-```sql
--- 开启结果缓存（SESSION 级别）
-SET cz.sql.enable.shortcut.result.cache = true;
-
--- 关闭结果缓存
-SET cz.sql.enable.shortcut.result.cache = false;
-```
-
-> 注意：当前默认未开启，需手动启用。
-
-## 缓存复用条件（同时满足才能命中）
-
-1. 查询中使用的表数据未发生变更
-2. 查询中不包含视图引用
-3. 新 SQL 与之前执行的 SQL 语法精确匹配
-4. 查询中不包含非确定性函数（如 `CURRENT_TIMESTAMP()`）或 UDF
-5. 之前的 Result Cache 未过期
-
-## 过期周期
-
-- 默认保留 **24 小时**
-- 24 小时内有查询复用，则额外延长 24 小时
-- 超过 24 小时无复用则清除
-
-## 约束与限制
-
-| 项目 | 限制 |
-|---|---|
-| 缓存保留周期 | 24 小时 |
-| 单工作空间最大缓存作业数 | 10 万 |
-| 缓存大小 | 无限制（≤10MB 存内存，>10MB 持久化到对象存储） |
-| 不支持 | 含非确定性函数、UDF 的查询 |
-
-## 验证是否命中缓存
-
-第二次执行相同查询后，在 Job Profile 的执行计划图中查看是否出现 `JOB RESULT REUSE` 标记。命中缓存的查询通常在 15ms 内返回。

package/bin/skills/clickzetta-query-optimizer/references/show-jobs.md

@@ -1,42 +0,0 @@
-# SHOW JOBS 参考
-
-> 来源：https://www.yunqi.tech/documents/show-jobs
-
-## 语法
-
-```sql
-SHOW JOBS [IN VCLUSTER vc_name] [LIKE 'pattern'] [WHERE <expr>] [LIMIT num];
-```
-
-## 参数说明
-
-- `IN VCLUSTER vc_name`：（可选）筛选指定计算集群下的作业
-- `WHERE <expr>`：（可选）按字段过滤，支持 SHOW JOBS 返回的所有字段
-- `LIMIT num`：（可选）限制返回数量，范围 1-10000
-- `LIKE 'pattern'`：（可选）按 job_id 模式匹配，支持 `%` 和 `_`
-
-默认显示最近 7 天内提交的任务，最多 10000 条。
-
-## 示例
-
-```sql
--- 查看执行时间超过 2 分钟的作业
-SHOW JOBS IN VCLUSTER default_ap WHERE execution_time > interval 2 minute;
-
--- 查看指定集群的所有作业
-SHOW JOBS IN VCLUSTER default_ap;
-
--- 限制返回 100 条
-SHOW JOBS LIMIT 100;
-
--- 查看指定集群最近 50 条
-SHOW JOBS IN VCLUSTER default_ap LIMIT 50;
-
--- 按 job_id 模式匹配
-SHOW JOBS LIKE 'job_2024%';
-```
-
-## 注意事项
-
-- 只能查看最近 7 天内的作业记录
-- 未指定 VCLUSTER 时显示所有集群的作业

package/bin/skills/clickzetta-realtime-sync-pipeline/eval_cases.jsonl

@@ -1,5 +0,0 @@
-{"case_id":"001","type":"should_call","user_input":"How do I create a single-table real-time sync task in Studio?","expected_skill":"clickzetta-realtime-sync-pipeline","expected_output_contains":["real-time sync","task_type","14"]}
-{"case_id":"002","type":"should_call","user_input":"How do I configure a single Kafka topic to sync to a Lakehouse table in real time?","expected_skill":"clickzetta-realtime-sync-pipeline","expected_output_contains":["Kafka","real-time sync"]}
-{"case_id":"003","type":"should_call","user_input":"What is the difference between single-table real-time sync and multi-table real-time sync?","expected_skill":"clickzetta-realtime-sync-pipeline","expected_output_contains":["single","multi","14","281"]}
-{"case_id":"004","type":"should_call","user_input":"How do I do MySQL single-table CDC real-time sync to Lakehouse?","expected_skill":"clickzetta-realtime-sync-pipeline","expected_output_contains":["MySQL","real-time sync","CDC"]}
-{"case_id":"005","type":"should_call","user_input":"Does a real-time sync task need scheduling configuration?","expected_skill":"clickzetta-realtime-sync-pipeline","expected_output_contains":["no scheduling","continuously running"]}

package/bin/skills/clickzetta-semantic-view/eval_cases.jsonl

@@ -1,12 +0,0 @@
-{"case_id":"001","type":"should_call","user_input":"How to create a semantic view","expected_skill":"clickzetta-semantic-view","expected_output_contains":["CREATE SEMANTIC VIEW"]}
-{"case_id":"002","type":"should_call","user_input":"How to query with semantic_view() function","expected_skill":"clickzetta-semantic-view","expected_output_contains":["semantic_view"]}
-{"case_id":"003","type":"should_call","user_input":"How to define dimensions and metrics in a semantic view","expected_skill":"clickzetta-semantic-view","expected_output_contains":["DIMENSIONS","METRICS"]}
-{"case_id":"004","type":"should_call","user_input":"How to use the semantic layer to unify metric definitions","expected_skill":"clickzetta-semantic-view","expected_output_contains":["semantic"]}
-{"case_id":"005","type":"should_call","user_input":"How to use SHOW SEMANTIC VIEWS","expected_skill":"clickzetta-semantic-view","expected_output_contains":["SHOW SEMANTIC VIEWS"]}
-{"case_id":"006","type":"should_call","user_input":"How to define FILTERS in a semantic view","expected_skill":"clickzetta-semantic-view","expected_output_contains":["FILTERS"]}
-{"case_id":"007","type":"should_call","user_input":"How to drop a semantic view with DROP SEMANTIC VIEW","expected_skill":"clickzetta-semantic-view","expected_output_contains":["DROP SEMANTIC VIEW"]}
-{"case_id":"008","type":"should_not_call","user_input":"How to create a regular view","forbidden_skill":"clickzetta-semantic-view"}
-{"case_id":"009","type":"should_not_call","user_input":"Help me build a data dashboard","forbidden_skill":"clickzetta-semantic-view"}
-{"case_id":"010","type":"should_not_call","user_input":"How to do data modeling","forbidden_skill":"clickzetta-semantic-view"}
-{"case_id":"011","type":"should_not_call","user_input":"How to connect Superset","forbidden_skill":"clickzetta-semantic-view"}
-{"case_id":"012","type":"should_not_call","user_input":"How to create a materialized view","forbidden_skill":"clickzetta-semantic-view"}

package/bin/skills/clickzetta-spark-flink-connector/eval_cases.jsonl

@@ -1,5 +0,0 @@
-{"case_id":"001","type":"should_call","user_input":"怎么用 Spark DataFrame 写入 ClickZetta Lakehouse？","expected_skill":"clickzetta-spark-flink-connector","expected_output_contains":["format(\"clickzetta\")","write"]}
-{"case_id":"002","type":"should_call","user_input":"Flink CDC 写入 ClickZetta 支持 UPDATE 和 DELETE 吗？","expected_skill":"clickzetta-spark-flink-connector","expected_output_contains":["igs-dynamic-table","CDC"]}
-{"case_id":"003","type":"should_call","user_input":"Spark Connector 支持写入主键表吗？","expected_skill":"clickzetta-spark-flink-connector","expected_output_contains":["不支持","Flink"]}
-{"case_id":"004","type":"should_call","user_input":"Flink 写入 ClickZetta 需要什么版本？","expected_skill":"clickzetta-spark-flink-connector","expected_output_contains":["1.15"]}
-{"case_id":"005","type":"should_call","user_input":"Flink 写入主键表的 sink.parallelism 有什么限制？","expected_skill":"clickzetta-spark-flink-connector","expected_output_contains":["1"]}

package/bin/skills/clickzetta-sql-migration/eval_cases.jsonl

@@ -1,10 +0,0 @@
-{"case_id":"001","type":"should_call","user_input":"How do I write Snowflake's IFF, ARRAY_SIZE, and LISTAGG in ClickZetta?","expected_skill":"clickzetta-sql-migration","expected_output_contains":["IF(","SIZE("]}
-{"case_id":"002","type":"should_call","user_input":"How to replace Databricks APPLY CHANGES INTO in ClickZetta?","expected_skill":"clickzetta-sql-migration","expected_output_contains":["MERGE INTO"]}
-{"case_id":"003","type":"should_call","user_input":"What are ClickZetta's implicit type conversion rules when migrating from Snowflake?","expected_skill":"clickzetta-sql-migration","expected_output_contains":["implicit","conversion"]}
-{"case_id":"004","type":"should_call","user_input":"How do I migrate a Snowflake VARIANT column to ClickZetta?","expected_skill":"clickzetta-sql-migration","expected_output_contains":["JSON"]}
-{"case_id":"005","type":"should_call","user_input":"Databricks ZORDER equivalent in ClickZetta","expected_skill":"clickzetta-sql-migration","expected_output_contains":["OPTIMIZE"]}
-{"case_id":"006","type":"should_call","user_input":"How to write Snowflake LATERAL FLATTEN in ClickZetta?","expected_skill":"clickzetta-sql-migration","expected_output_contains":["LATERAL VIEW EXPLODE"]}
-{"case_id":"007","type":"should_not_call","user_input":"How do I create a partitioned table in ClickZetta?","forbidden_skill":"clickzetta-sql-migration"}
-{"case_id":"008","type":"should_not_call","user_input":"What is the syntax for SELECT in ClickZetta?","forbidden_skill":"clickzetta-sql-migration"}
-{"case_id":"009","type":"should_not_call","user_input":"How to use window functions in ClickZetta?","forbidden_skill":"clickzetta-sql-migration"}
-{"case_id":"010","type":"should_not_call","user_input":"How do I create a Bloom Filter index?","forbidden_skill":"clickzetta-sql-migration"}

package/bin/skills/clickzetta-sql-pipeline-manager/eval_cases.jsonl

@@ -1,12 +0,0 @@
-{"case_id":"001","type":"should_call","user_input":"帮我创建一个动态表，每 5 分钟从 raw_events 聚合数据","expected_skill":"clickzetta-sql-pipeline-manager","expected_output_contains":["CREATE DYNAMIC TABLE","REFRESH INTERVAL"]}
-{"case_id":"002","type":"should_call","user_input":"怎么创建物化视图？","expected_skill":"clickzetta-sql-pipeline-manager","expected_output_contains":["CREATE MATERIALIZED VIEW"]}
-{"case_id":"003","type":"should_call","user_input":"创建一个 Table Stream 捕获 orders 表的变更","expected_skill":"clickzetta-sql-pipeline-manager","expected_output_contains":["CREATE TABLE STREAM"]}
-{"case_id":"004","type":"should_call","user_input":"怎么暂停动态表的刷新","expected_skill":"clickzetta-sql-pipeline-manager","expected_output_contains":["ALTER","SUSPEND"]}
-{"case_id":"005","type":"should_call","user_input":"怎么查看动态表的刷新历史","expected_skill":"clickzetta-sql-pipeline-manager","expected_output_contains":["SHOW DYNAMIC TABLE REFRESH HISTORY"]}
-{"case_id":"006","type":"should_call","user_input":"帮我设计一个 Medallion 架构的数据管道","expected_skill":"clickzetta-sql-pipeline-manager","expected_output_contains":["Bronze","Silver","Gold"]}
-{"case_id":"007","type":"should_call","user_input":"从 Kafka 持续导入数据到 Lakehouse 用什么方式","expected_skill":"clickzetta-sql-pipeline-manager","expected_output_contains":["Pipe","read_kafka"]}
-{"case_id":"008","type":"should_not_call","user_input":"帮我写一个 Node.js 后端","forbidden_skill":"clickzetta-sql-pipeline-manager"}
-{"case_id":"009","type":"should_not_call","user_input":"怎么创建用户和授权","forbidden_skill":"clickzetta-sql-pipeline-manager"}
-{"case_id":"010","type":"should_not_call","user_input":"Kubernetes 怎么部署","forbidden_skill":"clickzetta-sql-pipeline-manager"}
-{"case_id":"011","type":"should_not_call","user_input":"怎么连接 Superset","forbidden_skill":"clickzetta-sql-pipeline-manager"}
-{"case_id":"012","type":"should_not_call","user_input":"帮我优化一个慢查询","forbidden_skill":"clickzetta-sql-pipeline-manager"}

package/bin/skills/clickzetta-sql-pipeline-manager/evals/evals.json

@@ -1,166 +0,0 @@
-{
-  "skill_name": "clickzetta-sql-pipeline-manager",
-  "evals": [
-    {
-      "id": 1,
-      "prompt": "帮我创建一个动态表 dwd.orders_clean，每分钟刷新一次，从 ods.orders 表过滤 amount > 0 的数据，计算集群用 default_ap。",
-      "expected_output": "生成 CREATE OR REPLACE DYNAMIC TABLE dwd.orders_clean，包含 TARGET_LAG = '1 minutes'、VCLUSTER = default_ap、AS SELECT ... FROM ods.orders WHERE amount > 0",
-      "files": [],
-      "assertions": [
-        {"text": "包含 CREATE DYNAMIC TABLE 或 CREATE OR REPLACE DYNAMIC TABLE"},
-        {"text": "包含 dwd.orders_clean 作为表名"},
-        {"text": "包含 TARGET_LAG = '1 minutes' 或等效写法"},
-        {"text": "包含 VCLUSTER = default_ap"},
-        {"text": "包含 WHERE amount > 0 过滤条件"},
-        {"text": "包含 FROM ods.orders"}
-      ]
-    },
-    {
-      "id": 2,
-      "prompt": "我有一个动态表 dws.order_hourly 刷新失败了，怎么排查？",
-      "expected_output": "使用 SHOW DYNAMIC TABLE REFRESH HISTORY dws.order_hourly 查看失败详情，说明常见原因（SQL 报错、源表结构变更、集群不可用），提供排查步骤",
-      "files": [],
-      "assertions": [
-        {"text": "包含 SHOW DYNAMIC TABLE REFRESH HISTORY 命令"},
-        {"text": "包含 dws.order_hourly 表名"},
-        {"text": "提到至少一种常见失败原因（SQL报错/源表变更/集群不可用）"},
-        {"text": "提供具体的排查步骤或命令"}
-      ]
-    },
-    {
-      "id": 3,
-      "prompt": "我想从 Kafka 持续导入订单数据到 ods.orders 表，broker 是 kafka.prod.com:9092，topic 是 orders，数据是 JSON 格式。",
-      "expected_output": "生成 CREATE OR REPLACE PIPE 语句，使用 READ_KAFKA 函数，包含 KAFKA_BROKER、KAFKA_TOPIC、KAFKA_GROUP_ID、KAFKA_DATA_FORMAT = 'json'，以及 INSERT INTO ods.orders 的字段映射",
-      "files": [],
-      "assertions": [
-        {"text": "包含 CREATE PIPE 语句"},
-        {"text": "包含 READ_KAFKA 函数"},
-        {"text": "包含 kafka.prod.com:9092 作为 broker 地址"},
-        {"text": "包含 KAFKA_TOPIC 设置为 orders"},
-        {"text": "包含 KAFKA_DATA_FORMAT = 'json' 或等效写法"},
-        {"text": "包含 INSERT INTO ods.orders"}
-      ]
-    },
-    {
-      "id": 4,
-      "prompt": "我需要捕获 ods.orders 表的 UPDATE 和 DELETE 变更，然后同步到 dwd.orders_dim 表，怎么做？",
-      "expected_output": "先 CREATE TABLE STREAM ON TABLE ods.orders MODE = STANDARD，再创建 Dynamic Table 或用 MERGE INTO 消费 stream，说明 _change_type 字段含义（update_postimage / delete）",
-      "files": [],
-      "assertions": [
-        {"text": "包含 CREATE TABLE STREAM"},
-        {"text": "包含 ON TABLE ods.orders"},
-        {"text": "包含 MODE = STANDARD 或说明标准模式"},
-        {"text": "提到 _change_type 字段"},
-        {"text": "提到 update_postimage 或 delete 变更类型"},
-        {"text": "提供消费 stream 的方案（MERGE INTO 或 Dynamic Table）"}
-      ]
-    },
-    {
-      "id": 5,
-      "prompt": "帮我创建一个物化视图 dws.mv_daily_revenue，每小时自动刷新，统计每天每个 region 的 revenue 总和，数据来自 dwd.orders_clean。",
-      "expected_output": "生成 CREATE OR REPLACE MATERIALIZED VIEW，包含 REFRESH AUTO EVERY '1 hours'、VCLUSTER、AS SELECT DATE(created_at), region, SUM(amount) FROM dwd.orders_clean GROUP BY 1,2",
-      "files": [],
-      "assertions": [
-        {"text": "包含 CREATE MATERIALIZED VIEW"},
-        {"text": "包含 dws.mv_daily_revenue 作为视图名"},
-        {"text": "包含 REFRESH AUTO EVERY '1 hours' 或等效写法"},
-        {"text": "包含 FROM dwd.orders_clean"},
-        {"text": "包含 SUM 聚合 revenue 或 amount"},
-        {"text": "包含 GROUP BY 包括日期和 region"}
-      ]
-    },
-    {
-      "id": 6,
-      "prompt": "我要做一个三层 ETL：ODS 原始层 → DWD 清洗层 → DWS 聚合层，都用动态表，希望 DWD 每分钟刷新，DWS 跟着 DWD 刷新。怎么设计？",
-      "expected_output": "DWD 动态表设置 TARGET_LAG = '1 minutes'，DWS 动态表设置 TARGET_LAG = DOWNSTREAM，给出完整的两层 CREATE DYNAMIC TABLE 示例，说明 DOWNSTREAM 的含义",
-      "files": [],
-      "assertions": [
-        {"text": "包含两个 CREATE DYNAMIC TABLE 语句（DWD 和 DWS 层）"},
-        {"text": "DWD 层包含 TARGET_LAG = '1 minutes'"},
-        {"text": "DWS 层包含 TARGET_LAG = DOWNSTREAM"},
-        {"text": "解释 DOWNSTREAM 的含义（跟随上游刷新）"},
-        {"text": "DWS 层查询来自 DWD 层表"}
-      ]
-    },
-    {
-      "id": 7,
-      "prompt": "我需要暂停所有动态表的刷新，明天维护结束后再恢复，怎么操作？",
-      "expected_output": "使用 ALTER DYNAMIC TABLE <name> SUSPEND 暂停，ALTER DYNAMIC TABLE <name> RESUME 恢复，提示可先用 SHOW TABLES WHERE is_dynamic=true 列出所有动态表",
-      "files": [],
-      "assertions": [
-        {"text": "包含 ALTER DYNAMIC TABLE ... SUSPEND"},
-        {"text": "包含 ALTER DYNAMIC TABLE ... RESUME"},
-        {"text": "提到 SHOW TABLES WHERE is_dynamic=true 或等效方式列出动态表"}
-      ]
-    },
-    {
-      "id": 8,
-      "prompt": "动态表和物化视图有什么区别？我的场景是每天凌晨跑一次聚合报表，用哪个更合适？",
-      "expected_output": "说明动态表适合实时/增量场景，物化视图适合低频全量刷新；对于每天一次的报表场景推荐物化视图（REFRESH MANUAL 或 AUTO EVERY '1 days'），成本更低",
-      "files": [],
-      "assertions": [
-        {"text": "说明动态表适合实时或增量计算场景"},
-        {"text": "说明物化视图适合低频或手动刷新场景"},
-        {"text": "明确推荐物化视图用于每天一次的报表场景"},
-        {"text": "提到 REFRESH MANUAL 或 REFRESH AUTO EVERY '1 days' 等具体语法"}
-      ]
-    },
-    {
-      "id": 9,
-      "prompt": "帮我设计一个完整的数据管道：从 Kafka（broker: kafka.prod.com:9092，topic: user_events，JSON 格式）接入用户行为数据，字段有 user_id、event_type、page_id、ts（时间戳）。需要做到三层：ODS 原始层、DWD 清洗层（过滤无效 event_type）、DWS 聚合层（按天统计每个 page 的 UV 和 PV）。DWD 每 2 分钟刷新，DWS 跟随 DWD。",
-      "expected_output": "生成完整的端到端 SQL：1) ODS 建表，2) Kafka Pipe（使用 READ_KAFKA），3) DWD 动态表（TARGET_LAG='2 minutes'，VCLUSTER，过滤无效 event_type），4) DWS 动态表（TARGET_LAG=DOWNSTREAM，按天+page_id 聚合 UV/PV）",
-      "files": [],
-      "assertions": [
-        {"text": "包含 ODS 层建表语句（CREATE TABLE）"},
-        {"text": "包含 CREATE PIPE 使用 READ_KAFKA 从 kafka.prod.com:9092 摄入数据"},
-        {"text": "包含 DWD 层 CREATE DYNAMIC TABLE，TARGET_LAG = '2 minutes'，VCLUSTER"},
-        {"text": "包含 DWS 层 CREATE DYNAMIC TABLE，TARGET_LAG = DOWNSTREAM"},
-        {"text": "DWS 层包含 COUNT(DISTINCT user_id) 统计 UV 或 COUNT(*) 统计 PV"},
-        {"text": "DWS 层按日期和 page_id 分组（GROUP BY）"},
-        {"text": "包含验证命令（SHOW 或 DESC）"}
-      ]
-    },
-    {
-      "id": 10,
-      "prompt": "我有一个业务数据库的订单表会持续有 INSERT 和 UPDATE，已经同步到了 ClickZetta 的 ods.orders 表（字段：order_id, customer_id, amount, status, updated_at）。我需要构建一个完整的数仓管道：DWD 层保留最新状态的订单，DWS 层按天统计每个 status 的订单数和金额。要求近实时，延迟在 1 分钟以内。",
-      "expected_output": "使用 Table Stream（MODE=STANDARD）捕获 UPDATE 变更，DWD 动态表消费 stream 过滤 update_postimage，TARGET_LAG='1 minutes'，DWS 动态表 TARGET_LAG=DOWNSTREAM 做聚合",
-      "files": [],
-      "assertions": [
-        {"text": "包含 CREATE TABLE STREAM ON TABLE ods.orders，MODE = STANDARD"},
-        {"text": "DWD 层动态表消费 stream，过滤 _change_type IN ('insert', 'update_postimage')"},
-        {"text": "DWD 层包含 TARGET_LAG = '1 minutes'，VCLUSTER"},
-        {"text": "DWS 层包含 TARGET_LAG = DOWNSTREAM"},
-        {"text": "DWS 层按日期和 status 分组，包含 COUNT 和 SUM 聚合"},
-        {"text": "包含验证命令（SHOW DYNAMIC TABLE REFRESH HISTORY 或 SHOW TABLE STREAMS）"}
-      ]
-    },
-    {
-      "id": 12,
-      "prompt": "我想在 ClickZetta 上实现 Medallion Architecture，数据来自 S3（Volume 名称 s3_raw_vol，路径 /events/，JSON 格式），字段有 event_id、user_id、event_name、properties（JSON 嵌套）、created_at。Bronze 层直接落地原始数据，Silver 层做清洗（过滤 event_name 为空的记录，展开 properties 里的 page_url 和 referrer 字段），Gold 层按天统计每个 event_name 的次数和独立用户数。Silver 每 5 分钟刷新，Gold 跟随 Silver。",
-      "expected_output": "使用 bronze/silver/gold schema 命名（不映射到 ODS/DWD/DWS），生成完整 SQL：Bronze 建表 + OSS Pipe（COPY INTO FROM '@s3_raw_vol/events/'），Silver 动态表（TARGET_LAG='5 minutes'，过滤 event_name IS NULL，展开 properties:page_url 和 properties:referrer），Gold 动态表（TARGET_LAG=DOWNSTREAM，按天+event_name 聚合 COUNT 和 COUNT DISTINCT user_id）",
-      "files": [],
-      "assertions": [
-        {"text": "使用 bronze/silver/gold schema 命名（不使用 ods/dwd/dws）"},
-        {"text": "包含 CREATE PIPE 使用 COPY INTO FROM '@s3_raw_vol/events/'，FILE_FORMAT TYPE = 'json'"},
-        {"text": "Silver 层包含 CREATE DYNAMIC TABLE，TARGET_LAG = '5 minutes'，VCLUSTER，过滤 event_name IS NULL"},
-        {"text": "Silver 层展开 JSON 嵌套字段（properties:page_url 或 $1:page_url 等语法）"},
-        {"text": "Gold 层包含 CREATE DYNAMIC TABLE，TARGET_LAG = DOWNSTREAM"},
-        {"text": "Gold 层包含 COUNT(DISTINCT user_id) 和 COUNT(*) 或 COUNT(event_id)，按日期和 event_name 分组"}
-      ]
-    },
-    {
-      "id": 11,
-      "prompt": "我们有日志文件每小时上传到 OSS（Volume 名称 my_oss_vol，路径 /logs/app/，Parquet 格式），字段是 log_id、user_id、action、duration_ms、log_time。帮我构建完整的数据管道：自动导入到 ODS，然后做 DWD 清洗（过滤 duration_ms <= 0），最后做 DWS 层按天统计每个 action 的平均耗时和次数。不需要实时，每小时刷新一次就够。",
-      "expected_output": "使用对象存储 Pipe（COPY INTO from @my_oss_vol/logs/app/，TYPE=parquet），DWD 动态表 TARGET_LAG='1 hours'，DWS 动态表 TARGET_LAG=DOWNSTREAM，DWS 聚合 AVG(duration_ms) 和 COUNT(*)",
-      "files": [],
-      "assertions": [
-        {"text": "包含 CREATE PIPE 使用 COPY INTO FROM '@my_oss_vol/logs/app/'"},
-        {"text": "包含 FILE_FORMAT = (TYPE = 'parquet') 或等效写法"},
-        {"text": "DWD 层包含 CREATE DYNAMIC TABLE，TARGET_LAG = '1 hours'，VCLUSTER，过滤 duration_ms <= 0"},
-        {"text": "DWS 层包含 TARGET_LAG = DOWNSTREAM"},
-        {"text": "DWS 层包含 AVG(duration_ms) 统计平均耗时"},
-        {"text": "DWS 层按日期和 action 分组"}
-      ]
-    }
-  ]
-}

package/bin/skills/clickzetta-table-lineage/eval_cases.jsonl

@@ -1 +0,0 @@
-{"case_id":"001","type":"should_call","user_input":"Analyze table lineage over the past 7 days and generate a visualization page","expected_skill":"clickzetta-table-lineage"}

package/bin/skills/clickzetta-table-stream-pipeline/eval_cases.jsonl

@@ -1,5 +0,0 @@
-{"case_id":"001","type":"should_call","user_input":"怎么创建 Table Stream 捕获表的增量变更？","expected_skill":"clickzetta-table-stream-pipeline","expected_output_contains":["CREATE TABLE STREAM","change_tracking"]}
-{"case_id":"002","type":"should_call","user_input":"Table Stream 的 STANDARD 和 APPEND_ONLY 模式有什么区别？","expected_skill":"clickzetta-table-stream-pipeline","expected_output_contains":["STANDARD","APPEND_ONLY","INSERT"]}
-{"case_id":"003","type":"should_call","user_input":"Table Stream 消费后 offset 怎么管理？","expected_skill":"clickzetta-table-stream-pipeline","expected_output_contains":["offset"]}
-{"case_id":"004","type":"should_call","user_input":"怎么用 Table Stream + MERGE 实现幂等增量消费？","expected_skill":"clickzetta-table-stream-pipeline","expected_output_contains":["MERGE","幂等"]}
-{"case_id":"005","type":"should_call","user_input":"Table Stream 的 __change_type 元数据字段有哪些值？","expected_skill":"clickzetta-table-stream-pipeline","expected_output_contains":["__change_type","INSERT","UPDATE","DELETE"]}

package/bin/skills/clickzetta-vcluster-manager/LICENSE

@@ -1,16 +0,0 @@
-ClickZetta Skills License
-© 2026 Yunqi Inc. All rights reserved.
-LICENSE: Use of these materials (including all code, prompts, assets, files, and other components of these skills (collectively, "Skills")) is governed by your agreement with ClickZetta for the Service. If no separate agreement exists, use is governed by ClickZetta's Terms of Service (available at: https://yunqi.tech/documents/user-aggrement).
-Your applicable agreement is referred to as the "Agreement." "Service" is as defined in the Agreement.
-ADDITIONAL RESTRICTIONS: Notwithstanding anything in the Agreement to the contrary, you may not:
-
-Extract from the Service or retain copies of the Skills outside use with the Service;
-Reproduce or copy the Skills, except for temporary copies created automatically during authorized use of the Service;
-Create derivative works based on the Skills;
-Distribute, sublicense, or transfer the Skills to any third party;
-Make, offer to sell, sell, or import any inventions embodied in the Skills; nor,
-Reverse engineer, decompile, or disassemble the Skills.
-
-The receipt, viewing, or possession of the Skills does not convey or imply any license or right beyond those expressly granted above.
-Yunqi retains all rights, title, and interest in the Skills, including all copyrights, trademarks, patents, and all other applicable intellectual property rights.
-THE SKILLS ARE PROVIDED "AS IS," WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SKILLS OR THE USE OR OTHER DEALINGS IN THE SKILLS.