Skip to content

AGENTS.md

Latest commit

 

History

History
43 lines (34 loc) · 3.69 KB

AGENTS.md

File metadata and controls

43 lines (34 loc) · 3.69 KB
Raw

AI Platform Kit — agent guide

This repository is a set of skills for provisioning and operating Databricks platform infrastructure (workspaces, Unity Catalog, identity, networking, CI/CD) across Azure, AWS, and GCP. Each skill is a self-contained markdown file with instructions the agent follows.

How to use these skills

The skills live under .claude/skills/. Load only what the task needs: when a request matches a skill's description below, open that skill's SKILL.md and follow it. Cloud-specific companion files (AWS.md, AZURE.md, GCP.md, and their numbered topic files) are referenced from within each SKILL.md — read them based on the target cloud to avoid cross-cloud confusion.

Available skills

  • databricks-deployment-verification — MANDATORY post-deployment verification. After any workspace + Unity Catalog deployment, you MUST run all THREE compute paths against a UC table — classic cluster, serverless SQL warehouse, serverless notebook job. Skipping any of the three is incomplete work. Use whenever a workspace + UC has been freshly deployed or modified. Read .claude/skills/deployment-verification/SKILL.md when this applies.
  • databricks-identity-governance — Manage Databricks identity and governance. Use when the user asks to create groups, users, service principals, set up RBAC, manage permissions, workspace assignments, or configure access control. Read .claude/skills/identity-governance/SKILL.md when this applies.
  • databricks-platform-provisioning — Provision and test Databricks workspaces. Use when the user asks to create a workspace, set up a new environment, provision infrastructure, bootstrap Databricks, test a workspace, verify a deployment, or run validation checks against a Databricks workspace. Covers Azure, AWS, and GCP. Read .claude/skills/platform-provisioning/SKILL.md when this applies.
  • databricks-private-networking — Set up private networking for Databricks. Use when the user asks about private link, hub-spoke architecture, NCC (Network Connectivity Configuration), serverless connectivity to private resources, VPC endpoints, or private endpoints. Read .claude/skills/private-networking/SKILL.md when this applies.
  • databricks-unity-catalog-setup — Set up Unity Catalog on Databricks workspaces. Use when the user asks to configure Unity Catalog, create a metastore, set up catalogs, schemas, external locations, storage credentials, or configure data governance. Covers Azure, AWS, and GCP. Read .claude/skills/unity-catalog-setup/SKILL.md when this applies.
  • databricks-workspace-config — Configure Databricks workspace settings. Use when the user asks to create SQL warehouses, cluster policies, secret scopes, IP access lists, manage tokens, update workspace settings, or destroy infrastructure. Read .claude/skills/workspace-config/SKILL.md when this applies.

Operating notes

  • The kit writes Terraform and SDK/CLI calls tailored to the request — there are no rigid templates to fill in. Review every terraform plan before approving; this provisions real, billable cloud infrastructure.
  • Verify auth via cloud CLIs (aws sts get-caller-identity, az account show, gcloud auth list, databricks current-user me) rather than reading credential files.

Credential safety: Claude Code installs ship a PreToolUse hook that blocks reads of credential/state files (.databrickscfg, *.tfstate, cloud creds). Other agents have no equivalent hook — run them in a gated approval/sandbox mode so tool calls that touch those files require confirmation.