Skip to content
Get packs

Guide · Agent skill

How to use the RAG Enterprise Search on GKE and AlloyDB agent skill

Updated · written with AI from the project's own documentation and checked against our catalogue data

This is an agent skill, a folder of instructions that a coding agent such as Claude loads when your request matches its purpose. The RAG Enterprise Search on GKE and AlloyDB skill guides the agent through planning a retrieval-augmented generation (RAG) search system over a company's private documents on Google Cloud. After this guide you will know what the skill does, how to install it, how to phrase requests so it triggers, and what you need in place before using its output.

RAG Enterprise Search on GKE and AlloyDBgoogle/skills · Apache-2.0

Quick start

  1. 1Download the ZIP. Download the skill ZIP from this page.
  2. 2Unzip into the skills folder. Unzip into ~/.claude/skills/ for personal use, or into a project's .claude/skills/ folder.
  3. 3Or upload in the Claude apps. In the Claude apps, go to Settings → Capabilities → Skills and upload the ZIP.
  4. 4Check your Google Cloud setup. Confirm you have a project with AlloyDB, Cloud Storage and GKE access, plus gcloud CLI credentials if the agent will run commands.
  5. 5Describe your workload. Ask for RAG enterprise search on GKE and AlloyDB and describe your private documents.
  6. 6Answer the requirements questions. Answer the Phase 1 questions on functional and non-functional needs, then approve the technical decomposition.
  7. 7Review the architecture and scripts. Review the Phase 2 architecture and Phase 3 validation scripts before running anything against a real project.

What the skill makes the agent do

The skill runs a structured four-phase workflow. In Phase 1 the agent asks requirements questions about functional needs, plus security, reliability, cost, operational excellence, performance and sustainability, and it does not propose designs during this phase. Phase 2 turns the approved requirements into a component breakdown and architecture that uses an AlloyDB vector-enabled SQL database, Cloud Storage, and Google Kubernetes Engine (GKE) hosting an open model and an open-source inference framework.

Phase 3 covers validation: a plan, instructions and scripts to check the solution. Phase 4 consolidates everything into a solution guide. The skill's output template is organised around an executive summary, requirements and current state, and the architecture.

Install the skill

Download the ZIP for this skill from this page. Then choose one of two options.

For a personal install, unzip the file into ~/.claude/skills/. For a project-specific install, unzip it into the project's .claude/skills/ folder. Alternatively, in the Claude apps go to Settings, then Capabilities, then Skills, and upload the ZIP there.

  • Personal: ~/.claude/skills/
  • Project: .claude/skills/ inside your project
  • Claude apps: Settings → Capabilities → Skills, then upload the ZIP

Trigger the skill with natural requests

The agent loads the skill when your request matches its description: you need a vector-enabled SQL database as the store and index for embedding vectors, an open model with an open-source inference framework, and Kubernetes containers hosting the application components. Describe that situation in plain language and name the Google Cloud products if you know them.

  • "Help me plan a RAG search over our private contracts on Google Cloud using AlloyDB and GKE."
  • "Start the requirements discovery for an internal question-answering system over company documents."
  • "Generate the architecture and a validation plan for our RAG search, we have approved the requirements."

What you need before you start

The skill expects a Google Cloud project with access to AlloyDB, Cloud Storage and GKE. If the agent runs cloud commands, you also need the gcloud CLI or REST API or MCP tools, plus Google Cloud credentials for any commands it executes.

  • Google Cloud project with access to AlloyDB, Cloud Storage and GKE
  • gcloud CLI, or REST API or MCP tools, if the agent runs cloud commands
  • Google Cloud credentials for any commands the agent executes

Safety notes: scripts and network use

This skill runs scripts and uses the network. Before the agent runs any gcloud, REST API or MCP command, the skill requires it to follow its usage attribution instructions, which tag each gcloud command with an environment variable that identifies the skill and its version. The guidance and generated scripts still need human review and approval before you run them against real cloud projects.

Do not expect the skill to deploy and run production infrastructure unattended.

  • Review every generated script before running it.
  • Approve each phase before the agent moves on.
  • Run validation scripts against a non-production project first.

Tips for better results

Give concrete answers in Phase 1. Data types, ingestion frequency, latency targets and compliance rules shape the design. Resolve any contradictions the agent flags before asking for a design, because the skill will not produce architecture until they are settled. If you already have approved requirements or a confirmed architecture, say so, and the agent will skip the completed phases.

  • State data types (structured or unstructured) and ingestion frequency.
  • Name compliance and data residency rules up front.
  • Say which phases are already approved to skip them.

What people use it for

  • Planning an internal question-answering system over company policy documents.
  • Designing a self-hosted, SQL-based vector search stack on GKE and AlloyDB for engineering review.
  • Producing a requirements-to-design document for architects to review.
  • Drafting validation instructions and scripts for a proposed RAG architecture.
  • Compiling a consolidated solution guide from the approved phases.

Frequently asked questions

Is the RAG Enterprise Search on GKE and AlloyDB skill free?

The skill is published under the Apache-2.0 licence. Any Google Cloud resources it helps you plan or deploy are billed by Google Cloud, and this page does not state pricing.

What do I need to use this skill?

You need a Google Cloud project with access to AlloyDB, Cloud Storage and GKE. If the agent runs cloud commands, you also need the gcloud CLI or REST API or MCP tools and Google Cloud credentials.

How do I install the agent skill?

Download the ZIP from this page and unzip it into ~/.claude/skills/ for personal use or a project's .claude/skills/ folder. You can also upload it in the Claude apps under Settings → Capabilities → Skills.

Does the skill deploy my infrastructure?

Not unattended. The skill plans, produces guidance and can draft validation scripts, but the guidance and scripts need human review and approval before you run them against real cloud projects.

When should I use a different tool instead?

Use something else if you want a fully managed RAG or SaaS search service, or if you need a non-SQL vector database. The skill is explicitly scoped to SQL-based vector storage and says not to use it for those cases.

Why won't the agent give me an architecture right away?

In Phase 1 the agent is required to ask requirements questions without proposing designs. It only generates a technical decomposition after ambiguities are resolved, and it waits for your approval before moving to architecture.

New agent skills, repos and AI tools — every week

The best new skills and open-source projects, the fastest-growing repos, and one free prompt. Five minutes to read. Unsubscribe anytime.

Our verdict

Choose this skill if you are an engineering team or architect planning a self-hosted, SQL-based vector search system over private documents on Google Cloud and want a disciplined requirements-first process. Pick a managed RAG or SaaS search service instead if you want someone else to run the infrastructure, or a non-SQL vector database if that is a hard requirement.

Download the skill