AgentSkillsCN

setup

为分析技能配置 Snowflake 连接——在克隆仓库后,务必先运行此步骤。

SKILL.md
--- frontmatter
name: setup
description: Configure your Snowflake connection for the analytics skills - run this first after cloning the repo

Setup — Configure Snowflake Connection

You are a setup assistant for the CS Usage Analytics toolkit. Help the user configure their Snowflake environment so all analytics skills work correctly.

When to Run

  • First time using the toolkit after cloning the repo
  • When switching Snowflake environments
  • When CLAUDE.md still has YOUR_DATABASE / YOUR_SCHEMA placeholders

Parameter Collection

Ask the user for:

  1. Database (required): "What is your Snowflake database name? (e.g., LANDING, MDLH_AWS_ATLANVC_CONTEXT_STORE)"
  2. Schema (required): "What is the schema containing the PAGES, TRACKS, and USERS tables? (e.g., FRONTEND_PROD, usage_analytics)"

Execution

Once you have both values:

  1. Update CLAUDE.md — Edit ~/atlan-usage-analytics/CLAUDE.md and replace the Configuration table values:

    • Replace YOUR_DATABASE (or whatever the current DATABASE value is) with the user's database name
    • Replace YOUR_SCHEMA (or whatever the current SCHEMA value is) with the user's schema name
    • Remove the — run /setup to configure suffixes from the Description column if present
  2. Validate the connection — Run a quick test query to confirm the config works:

    sql
    SELECT COUNT(*) AS row_count FROM <DATABASE>.<SCHEMA>.PAGES WHERE TIMESTAMP >= DATEADD('day', -7, CURRENT_TIMESTAMP())
    

    Execute via mcp__snowflake__run_snowflake_query.

  3. Report results:

    • If the query succeeds: "Connected! Found X page views in the last 7 days. Your analytics skills are ready to use."
    • If it fails: Show the error and ask the user to double-check their database/schema names.

Post-Setup

After successful configuration, suggest next steps:

  • /discover domains — See which customer domains are available
  • /health <domain> — Run a health check on a specific customer
  • /features <domain> — See what features a customer is using
  • Type list all analyses available to see everything that's available