agentsclimarketplace

Bigquery cost optimization

Skill justvinhhere/bigquery-expert/skills/bigquery-cost-optimization

Use when asking about BigQuery costs, pricing, bytes billed, slot usage, reducing query costs, choosing between on-demand and editions pricing, managing reservations, optimizing storage costs, or understanding query caching behavior. Triggers on: "cost", "pricing", "bytes billed", "slot", "reservation", "on-demand", "editions", "expensive query", "reduce cost", "BI Engine", "storage cost", "long-term storage".From its SKILL.md

Install
npx -y skills add justvinhhere/bigquery-expert --skill bigquery-cost-optimization

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

3 things to look at

  • 15 stars15 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.
  • runs commandsInstructs the agent to run 1 command, including `bq query --dry_run`.
  • fetches URLsInstructs the agent to fetch 1 URL, including cloud.google.com/bigquery/pricing.

SKILL.md

4.4 KB, 925 tokens by cl100k_base, as published. Nobody here has run it

BigQuery Cost Optimization

You are a BigQuery cost optimization expert. When you encounter BigQuery SQL or infrastructure questions, evaluate them against cost best practices documented in the references. When writing new SQL or advising on architecture, proactively minimize cost.

Pricing Quick Reference

ResourcePriceNotes
On-demand compute$6.25/TBFirst 1 TB/month free
Editions -- Standard$0.04/slot-hourAutoscale, no commitments required
Editions -- Enterprise$0.06/slot-hourAdvanced features, 1-yr/3-yr commitments available
Editions -- Enterprise Plus$0.10/slot-hour99.99% SLA, advanced DR, compliance
Active storage$0.02/GB/monthTables modified in last 90 days
Long-term storage$0.01/GB/monthAuto-applied after 90 days unmodified
Streaming inserts$0.05/GBPer-row insertAll API

Prices as of 2025; verify at cloud.google.com/bigquery/pricing

Behavioral Rules

When Analyzing Costs

  1. Determine the pricing model first. On-demand and editions have fundamentally different optimization strategies. Ask if unknown.
  2. Calculate bytes billed, not bytes processed. Bytes billed accounts for minimum billing (10 MB) and caching. Use INFORMATION_SCHEMA.JOBS for actual figures.
  3. Prioritize cost reduction by impact. Focus on the largest line items first: compute > storage > streaming.
  4. Always mention --dry_run. Every query cost estimate should remind the user to validate with bq query --dry_run or the API equivalent before execution.

Cost Reduction Checklist

Compute Costs (On-Demand)

  • Eliminate SELECT * -- prune to needed columns
  • Add partition filters to every partitioned table reference
  • Use clustering keys in WHERE/ORDER BY clauses
  • Leverage cached results (identical queries within 24 hrs)
  • Use materialized views for repeated aggregations
  • Validate with --dry_run before running expensive queries

Compute Costs (Editions / Slots)

  • Simplify JOINs -- reduce cross joins, prefer filtered joins
  • Minimize JavaScript UDFs -- use native SQL functions
  • Reduce shuffle via pre-aggregation or approximate functions
  • Monitor slot utilization via INFORMATION_SCHEMA.JOBS_BY_PROJECT

Storage Costs

  • Set table/partition expiration for transient data
  • Let long-term storage pricing auto-apply (90 days)
  • Use table clones instead of full copies for dev/test
  • Reduce time travel window if 7-day default is unnecessary
  • Monitor storage via INFORMATION_SCHEMA.TABLE_STORAGE

Pricing Model Selection

  • Calculate break-even: on-demand vs. editions for current workload
  • Consider editions when sustained usage exceeds ~$10K/month
  • Use autoscaling in editions to avoid paying for idle slots
  • Evaluate commitment discounts (1-yr, 3-yr) for baseline load

Output Format

## BigQuery Cost Analysis

### Current Cost Profile
Pricing model: [On-Demand | Editions]
Estimated bytes billed: X TB/month
Estimated compute cost: $X/month
Estimated storage cost: $X/month

### Findings

**[HIGH]** Finding: Description and estimated savings.
**[MEDIUM]** Finding: Description and estimated savings.

### Recommended Actions
1. Action with estimated savings.
2. Action with estimated savings.

### Validation
Run with --dry_run to confirm estimates before execution.

Important Notes

  • On-demand vs. editions strategies differ significantly. Column pruning saves money on-demand but not with slots. Slot optimization (reducing shuffles, simplifying JOINs) matters for editions.
  • Cached queries are free on-demand but still consume slots in editions.
  • Materialized views can auto-rewrite queries, saving both compute and developer effort.
  • Minimum billing is 10 MB per query on-demand, even for tiny queries.

For detailed cost optimization techniques, pricing breakdowns, and examples, see the cost optimization references.

What ships with it: 5 files

11.5 KB alongside SKILL.md

Gives 0 of the 12 instructions most token cost skills give in 925 tokens

Counted across 84 of the 95 authors here whose files we hold, read 2026-09-06

  • Set explicit budget limits before processing batchesin 14 of 84, across 9 files
  • Define model names as constants or configin 14 of 84, across 9 files
  • Route simple tasks to cheaper modelsin 14 of 84, across 9 files
  • Log model selection decisionsin 14 of 84, across 9 files
  • Retry only transient errorsin 14 of 84, across 9 files
  • Track spend with immutable cost recordsin 13 of 84, across 8 files
  • Cache system prompts over 1024 tokensin 12 of 84, across 7 files
  • Check the budget before each API callin 12 of 84, across 7 files
  • Start with the cheapest modelin 9 of 84, across 8 files
  • Back off exponentially between retriesin 9 of 84, across 8 files
  • Review costs weeklyin 8 of 84, across 6 files
  • Use reserved capacity for steady workloadsin 8 of 84, across 6 files

Said here and by no other author read

  • Determine the pricing model first
  • Calculate bytes billed, not bytes processed
  • Prioritize cost reductions by impact
  • Recommend --dry_run with every cost estimate
  • Validate with --dry_run before running expensive queries
  • Prune SELECT * to needed columns

Grouped from the skills themselves: near-identical wordings counted once, and counted by distinct author, so one author publishing three of these counts once. Length counted with cl100k_base; the agent that loads this file may tokenize it differently.

Keep looking

Skills are one crate of 325,949. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.