Claude-skill-registry bq-query

Design and execute BigQuery queries. Use for schema exploration, writing SQL from requirements, running queries, checking costs, or validating syntax.

install

source · Clone the upstream repo

git clone https://github.com/majiayu000/claude-skill-registry

Claude Code · Install into ~/.claude/skills/

T=$(mktemp -d) && git clone --depth=1 https://github.com/majiayu000/claude-skill-registry "$T" && mkdir -p ~/.claude/skills && cp -r "$T/skills/data/bq-query" ~/.claude/skills/majiayu000-claude-skill-registry-bq-query && rm -rf "$T"

manifest: skills/data/bq-query/SKILL.md

BigQuery Query

Rules: Follow coding-standards for SQL naming and readability.

Prerequisites

Check gcloud configuration before running queries:

gcloud config get-value project

If authentication error: prompt user to run
```
gcloud auth login
```
, then resume
If project unset: prompt user to run
```
gcloud config set project <PROJECT_ID>
```

Workflow

Clarify requirements: Understand what data is needed and why. If requirements are already in context, proceed to next step.

Understand schema: Explore available datasets. If schema is already in context, proceed to next step.

bq ls project:dataset                           # List tables
bq show --schema project:dataset.table          # Show table schema

Design query: Write SQL based on requirements and schema.
- Use CTEs for readability
- Use fully-qualified table names:
```
project.dataset.table
```
- Specify exact date ranges to limit scanned data
- Filter partitioned tables by partition key
- Avoid correlated subqueries (use JOINs/CTEs)
- Filter early with CTEs before joining large tables
- Use
```
LIMIT
```
  for exploration queries

Dry run: Validate syntax and estimate cost

bq query --use_legacy_sql=false --dry_run "SELECT ..."

Cost: ~$5/TB. If >2GB, ask user before executing.

Execute: Run after successful dry run, once any required user confirmation has been given

bq query --use_legacy_sql=false --format=csv "SELECT ..."