菜单
Validate migrated data between source and Snowflake using cloud validation (SPCS). Setup, run, background Monitor (or poll fallback), and end-of-run summary. Triggers: validate data, validate tables, data validation, check data.
默认使用会先检查来源的 Prompt；你也可以切换为直接命令，或下载本地副本。
决定是否安装前，请先阅读 SKILL.md，以及 SkillsMP 当前展示的配套文件。
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
直接命令不会经过审查 Prompt；运行前请先检查来源。
npx skills add https://github.com/Snowflake-Labs/cortex-code-migrations --skill validate-objects命令会保持在同一行。复制前请横向滚动并检查完整内容。
想先保存到本地？可下载 SkillsMP 当前能够提供的文件。
Shared infrastructure setup for data migration and data validation — prerequisites, compute pool, worker config, and source database/schema capture.
Setup, run, and report on cloud data migration — workflow YAML, migrate_data run, background Monitor (or poll fallback), and end-of-run summary for the user.
Core migration loop for a single stored procedure, function, or BTEQ script. Iterates deploy -> test -> diagnose -> fix until Snowflake output matches source.
基于 SOC 职业分类
正在显示 SKILL.md
| name | validate-objects |
| description | Validate migrated data between source and Snowflake using cloud validation (SPCS). Setup, run, background Monitor (or poll fallback), and end-of-run summary. Triggers: validate data, validate tables, data validation, check data. |
Tell the user:
Data Validation — I'll compare your migrated data against the source to verify row counts, schema matches, and data integrity.
Call configure() to retrieve the current configuration.
Check the returned values for snowflake_connection, source_connection, and snowflake_database:
configure with the updated values.configure with all of them.Ensure the shared infrastructure is up before validating: run data_infrastructure(mode="up") — pass compute_pool="<POOL>" for SPCS (ask the user for the pool name) or omit it for a local orchestrator.
Load actions/validate_tables.md (Steps 4–6: background Monitor or poll fallback, error-first report, teardown).
The error-first data validation report (Result + Workflow, Errors, Suggested fixes) is produced in actions/validate_tables.md Step 5 via job_status(job_id, details=true). Do not substitute migration_status() for that report.
After validate_tables.md completes (including the summary and teardown offer), optionally call migration_status() for wave-level context only:
configure() call will auto-advance to the next wave.migration_status() shows no remaining waves (all-waves-done), invoke ../data-infrastructure/teardown/SKILL.md unconditionally before returning to the parent — in addition to the per-wave teardown prompt in validate_tables.md Step 6 when applicable.