Build an end-to-end pipeline that ingests Cloudflare HTTP request logs from R2 into Snowflake and exposes them through a natural-language queryable Cortex Agent.
- Raw table — VARIANT storage of Cloudflare HTTP request JSON
- Parsed view — ~75 typed columns (timestamps, geo, cache, bot, WAF, origin, Workers)
- Semantic view — 28 facts, 42 dimensions, 16 pre-built metrics with AI hints
- Cortex Agent — ask questions in plain English about traffic, errors, bots, cache, security, and more
- Optional scheduled task — weekly refresh from R2
- Snowflake account with ACCOUNTADMIN (or CREATE DATABASE + CREATE AGENT privileges)
- Cloudflare Logpush configured to push HTTP request logs to an R2 bucket
- R2 API credentials (access key + secret key)
This sample is designed as a Cortex Code skill. Install the skill and invoke it — it will walk you through configuration, infrastructure creation, data loading, semantic view setup, and agent creation step by step.
See SKILL.md for the full workflow.