Skip to content
Get packs

Guide · Agent skill

How to use the Bigtable Basics agent skill

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

The Bigtable Basics skill is a set of instructions that teaches an AI agent how to work with Google Cloud Bigtable: designing row keys, creating instances, tables and column families, writing Bigtable SQL, and writing Java, Go or Python client code. This guide explains what a skill is, how to install this one, how to ask for help in plain language, and what to check before letting the agent touch a live database. You do not need prior experience with agent skills to follow it.

Bigtable Basicsgoogle/skills · Apache-2.0

Quick start

  1. 1Download and unzip. Download the bigtable-basics ZIP from this page and unzip it so the bigtable-basics folder contains SKILL.md.
  2. 2Place or upload the skill. Copy the folder to ~/.claude/skills/ or a project's .claude/skills/, or upload it under Settings → Capabilities → Skills in the Claude apps.
  3. 3Install the prerequisites. Install the Google Cloud SDK (gcloud) and the cbt CLI, and make sure you have a Bigtable-enabled project with IAM access if you plan to work live.
  4. 4Set up cbt. Set BIGTABLE_PROJECT and BIGTABLE_INSTANCE, then write the .cbtrc file with the echo commands shown above.
  5. 5Ask in plain language. Describe your Bigtable task and name Bigtable, for example a row key design or a SQL query.
  6. 6Review and approve. Read each command the agent proposes and give explicit confirmation before any change to a non-emulator database.

What the Bigtable Basics skill makes the agent do

Once installed, the agent gets Google Bigtable specific guidance. It is built for administering and developing with Bigtable, and it covers provisioning instances and tables, designing performant schemas, and querying data.

The skill tells the agent to watch for row key design, since Bigtable efficiency depends on it, and to warn about full table scans. When diagnosing slow performance or hotspotting, it points the agent to Key Visualizer in the Cloud Console first, then to the hot-tablets tool, table stats in the gcloud CLI, and the include-stats=full option under cbt read.

  • Row key and column family design, including avoiding hotspots
  • Provisioning with gcloud (instances, clusters, app profiles, backups, IAM) and cbt (table and column family updates, reading and writing data)
  • Bigtable SQL queries for reads and aggregations
  • Java, Go, Python client patterns, with Java or Go suggested for production
  • Performance diagnosis that starts with Key Visualizer

Install the skill (ZIP download or Claude apps)

Download the ZIP for this skill from this page and unzip it. The folder it contains is named bigtable-basics and includes SKILL.md plus a references folder and an assets folder.

For Claude Code, place the unzipped folder in ~/.claude/skills/ to make it available across your projects, or in a project's .claude/skills/ folder to limit it to that project.

In the Claude apps, you can upload the skill under Settings → Capabilities → Skills instead.

  • Personal install: ~/.claude/skills/bigtable-basics
  • Project install: .claude/skills/bigtable-basics inside your project

Set up what the skill needs

The skill does not run scripts, but it does use the network. For provisioning and data commands you need the Google Cloud SDK (gcloud) and the cbt CLI. For live use you need a Google Cloud project with Bigtable access and suitable IAM permissions. If the agent generates application code, you also need the Bigtable client libraries for Java, Go or Python.

The skill's cbt setup writes a .cbtrc file that stores your project and instance. Set the BIGTABLE_PROJECT and BIGTABLE_INSTANCE variables first so the values are filled in correctly.

  • Google Cloud SDK (gcloud) and cbt CLI
  • A Google Cloud project with Bigtable access and IAM permissions for live work
  • Bigtable client libraries in Java, Go or Python, only if generating code
echo project = ${BIGTABLE_PROJECT} > ~/.cbtrc echo instance = ${BIGTABLE_INSTANCE} >> ~/.cbtrc

Trigger the skill with natural requests

You do not need a special command. Describe the task in plain words that mention Bigtable, and the agent should load the skill when the request matches its description. The skill's description covers row key design, column families, SQL queries, client code, performance and hotspotting, and provisioning with gcloud or cbt.

Name Bigtable explicitly in your request. The skill says not to use it for generic Cloud SQL administration, so a request about a relational database will not match.

  • "Design a row key for a Bigtable table storing sensor readings so writes don't hotspot."
  • "Write a Bigtable SQL query that counts events per user prefix in the events table."
  • "Why is my Bigtable table slow? Help me diagnose hotspotting."
  • "Show me Go code to write rows to Bigtable with the client library."

Confirm changes and read commands before running them

The skill contains a safety rule: the agent must get explicit confirmation from you before making any change to a non-emulator database, and it must mention this requirement when giving commands that change structure or data. Read every command the agent proposes and approve it yourself.

The skill also says to prefer Terraform for production schema changes to prevent accidental data loss. For manual cbt changes, the agent is told to list the existing column families and GC policies first.

  • Review each create, update or delete command before approving it
  • Use Terraform for production schema changes where possible
  • Test on an emulator or non-production instance first
cbt ls {table} cbt createfamily {table} {family} cbt setgcpolicy {table} {family} "maxversions=5 AND maxage=30d"

Check the attribution rule and network use

Before running any gcloud, REST API or MCP command, the skill requires the agent to read its attribution reference. Terminal gcloud commands are prefixed with a CLOUDSDK_METRICS_REQUEST_ATTRIBUTION environment variable that identifies the skill and its version. The skill says this variable is set inline on each command and not saved with gcloud config set. The tag carries no user data, but you can see it in every terminal command the agent runs.

Because the skill uses the network, commands run against your Google Cloud project. The skill does not run scripts itself.

  • Runs scripts: no
  • Uses network: yes, for Google Cloud commands and API calls
  • Tagged gcloud commands are not written into generated scripts, Terraform or user code

What people use it for

  • Designing a row key for time-series data so writes spread across tablets instead of hotspotting
  • Writing a Bigtable SQL query that uses a _key range or prefix to avoid unbounded scans
  • Diagnosing slow reads by starting with Key Visualizer, then checking hot tablets and table stats
  • Creating or updating a column family and its GC policy with cbt after checking the current state
  • Generating Java, Go or Python client code for reading and writing rows

Frequently asked questions

How do I install the Bigtable Basics agent skill?

Download the ZIP from this page and unzip it. Place the bigtable-basics folder in ~/.claude/skills/ for personal use or in a project's .claude/skills/ folder, or upload it in the Claude apps under Settings → Capabilities → Skills.

What does the bigtable-basics skill require?

It needs the Google Cloud SDK (gcloud) and the cbt CLI for provisioning and data commands, a Google Cloud project with Bigtable access and IAM permissions for live use, and Bigtable client libraries in Java, Go or Python if the agent generates code. It does not run scripts but does use the network.

Is the Bigtable Basics skill free to use?

The source for this skill is licensed under Apache-2.0. The skill's pricing for Google Cloud usage is not covered here, so check Google Cloud's own pricing for Bigtable and any tools you use.

Will the agent change my live Bigtable tables without asking?

No, the skill instructs the agent to obtain explicit user confirmation before making non-emulator database changes. You should still read each command before approving it, since the skill relies on the agent following that rule.

Can this skill help with Cloud SQL or other relational databases?

No. The skill says it does not cover generic Cloud SQL administration. Use it for Bigtable tasks such as row keys, cbt and gcloud commands for Bigtable, Bigtable SQL and client code.

Why does my Bigtable table have slow queries or hotspots?

The skill's diagnostic path starts with Key Visualizer in the Cloud Console, which shows access patterns across row keys. It then suggests the hot-tablets tool, table stats in the gcloud CLI, and the include-stats=full option under cbt read.

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

This skill suits engineers designing or tuning Bigtable schemas, data teams running SQL or cbt queries, and developers who want Java, Go or Python client patterns. It is not a good fit for generic relational database work, or for anyone who wants an agent to change production tables without reviewing each command.

Download the skill