Skip to content

alirezarezvani/claude-skills / engineering/skills/cloud-finops

cloud-finops

Verified

Use for any cloud cost question on AWS, Azure or Google Cloud - a bill that jumped or one service that spiked (NAT gateway, egress, storage), cost anomaly alert thresholds, untagged or unallocated spend and tag cleanup for showback or chargeback, Savings Plans, Reserved Instances or committed use discounts (utilization, coverage, unused commitment, effective savings rate, whether to buy more), or whether cost per order or per customer got worse. Reads a FOCUS billing export with stdlib scripts.

Cloud FinOps

Portability: 3 stdlib Python scripts plus a shared loader. Input is a CSV you export; no cloud credentials, no API calls. Definitions verified 2026-11-17 against FOCUS v1.4 and the FinOps Foundation.

A cloud bill answers "how much"; it rarely answers "whose", "was the discount worth it" or "is this spike real". FOCUS, the open billing schema that AWS, Azure and Google Cloud all export, makes the three questions computable the same way on every provider. This skill computes them and refuses the common shortcuts: allocating by BilledCost, measuring coverage in discounted dollars, buying commitments while existing ones sit idle, and alerting on percentages alone.

Workflow

# 0 - Export FOCUS: AWS Data Exports (FOCUS 1.x), Azure Cost Management export (FOCUS), GCP BigQuery FOCUS view
# 1 - Who owns the spend?
python scripts/allocation_check.py --input focus.csv --required-tags team,env,cost-center
# 2 - Are the commitments earning their keep? Should we buy more?
python scripts/commitment_check.py --input focus.csv --eligible Compute
# 3 - What changed, and is it the business or the bill?
python scripts/cost_anomaly.py --input focus.csv --units orders.csv

Step 1 reports the share of EffectiveCost carrying every required tag, the largest unallocated resources, and tag values that split one owner (Prod/prod). Cost with no ResourceId (shared charges, unused commitment) needs a split rule, not a tag.

Step 2 computes utilization, coverage in on-demand equivalent (ListCost), and the effective savings rate with unused commitment counted as negative savings. It names under-used commitments, says "fix utilization first" when coverage is low and commitments are idle, and gives the lowest daily uncovered spend as the size a new commitment would keep fully used.

Step 3 flags a group-day only when it beats a robust z-score (median and MAD), a relative floor and a money floor together, and compares cost per business unit between the two halves of the period. Under 14 days it refuses.

Hard rules

  • EffectiveCost for ownership and trends; BilledCost only for the invoice. Purchases are spread by EffectiveCost.
  • Never recommend a commitment purchase while utilization is under target. Move workloads onto what is paid for first.
  • Size new commitments to the floor, not the average. Above the floor is a growth bet; say it is one.
  • Thresholds and targets are policy. State every default you used (--target, --min-coverage, -k, --min-abs).
  • A total that rises with the business is not an anomaly. Check unit cost before calling a spike waste.
  • Read-only. Never change tags, buy or exchange commitments, or delete resources; produce the recommendation and its numbers.

Scripts

ScriptRoleExit codes
scripts/allocation_check.pyAllocation coverage, spend by category and account, unallocated resources, tag hygiene0 at target · 2 findings · 3 bad input
scripts/commitment_check.pyUtilization, coverage, ESR, per-commitment waste, sizing floor0 healthy · 2 findings · 3 bad input
scripts/cost_anomaly.pyRobust daily anomalies by service or account, unit-cost drift0 none · 2 findings · 3 bad input

All support --help, --sample and --output json. scripts/focus_common.py is the shared loader.

References and assets

finishing-a-development-branchobra/superpowersUse when implementation is complete, all tests pass, and you need to decide how to integrate the workStars: 322kUpdated Claude CodeCodexVerifiedInstall
systematic-debuggingobra/superpowersUse when encountering any bug, test failure, or unexpected behavior, before proposing fixesStars: 322kUpdated Claude CodeCodexVerifiedInstall
test-driven-developmentobra/superpowersUse when implementing any feature or bugfix, before writing implementation codeStars: 322kUpdated Claude CodeCodexVerifiedInstall
skill-creatoranthropics/skillsCreate new skills, modify and improve existing skills, and measure skill performance. Use when users want to create a skill from scratch, edit, or optimize an existing skill, run evals to test a skill, benchmark skill performance with variance analysis, or optimize a skill's description for better triggering accuracy.Stars: 198kUpdated Claude CodeCodexVerifiedInstall