菜单
Use when answering questions about data in a warehouse (BigQuery, Snowflake, etc.) and you have direct SQL access (CLI, driver, or console). Discover metadata and run knowledge search through the warehouse-native AGENTS schema instead of INFORMATION_SCHEMA.
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。
基于 SOC 职业分类
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/dbt-labs/agents_schema --skill agents-schema-search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。
正在显示 SKILL.md
| name | agents-schema-search |
| description | Use when answering questions about data in a warehouse (BigQuery, Snowflake, etc.) and you have direct SQL access (CLI, driver, or console). Discover metadata and run knowledge search through the warehouse-native AGENTS schema instead of INFORMATION_SCHEMA. |
Before answering a warehouse-data question, read the warehouse AGENTS schema.
AGENTS.ROOT is the entry point for the warehouse-native Agents Schema. It tells AI agents which providers have published metadata and how to use their provider-specific tables.
Use AGENTS as the warehouse metadata and discovery surface. Treat it like an agent-oriented alternative to INFORMATION_SCHEMA: inspect AGENTS.ROOT and the provider-specific AGENTS.* tables instead of querying INFORMATION_SCHEMA directly. Only fall back to INFORMATION_SCHEMA when the AGENTS content explicitly says to verify live metadata, or when required metadata is missing from AGENTS.
Use your connection's SQL dialect (e.g. GoogleSQL on BigQuery), and quote schema and table identifiers using that dialect. Run read-only SELECT statements only.
Required first query:
SELECT * FROM AGENTS.ROOT ORDER BY provider, key;
Then follow the provider guidance returned in the content column.
If no rows are returned, or the AGENTS schema or the AGENTS.ROOT table is not found, tell the user that their AGENTS schema is not set up yet.