agentleFS
harshgoyal1harshgoyal1/cloud-architect-skills/skills/finops-review/SKILL.md

finops-review

Skill1 starsChanged 3 months agoMITView on GitHub

FinOps review of a cloud estate or workload — cost visibility, allocation, waste, commitment coverage, unit economics and the practices that keep spend under control. Use when the user asks to reduce cloud cost, review a bill, find waste, plan reserved instances or savings plans, set up showback/chargeback, or asks why cloud spend is rising.

The file

We split this skill into the rules it gives an agent

2 rules. About 0 are common, 2 are distinctive, and 0 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈2

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 0

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Follow the FinOps lifecycle: Inform → Optimize → Operate.

    line 8
  2. Distinctive:

    Never recommend a commitment purchase before waste and rightsizing — you would be locking in the current inefficiency for one to three …

    line 81

≈ “Common” is an estimate: at least 1 in 5 of the 372,429 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

FinOps Review

Follow the FinOps lifecycle: Inform → Optimize → Operate. Skipping to Optimize is the most common failure — you cannot cut what you cannot attribute.

Method

  1. Get the actual data: cost export / CUR, utilisation metrics, tag coverage.
  2. Establish the baseline: total, trend, top 10 services, top 10 resource groups/accounts.
  3. Work the phases in order. Report savings as annualised, with effort and risk.
  4. Separate waste (delete, zero risk) from optimisation (change, some risk) from commitment (pre-pay, financial risk).

Inform — visibility and allocation

  • Tag coverage percentage; untagged spend is the first number to quote
  • Mandatory tags enforced by policy: owner, cost centre, environment, application
  • Allocation model: showback or chargeback; shared cost handling (network, platform, support)
  • Anomaly detection configured; who receives it
  • Unit economics: cost per customer / transaction / tenant — the only metric leadership acts on
  • Forecast vs actual variance for the last 3 months

Optimize — waste (delete first, it is free)

  • Unattached disks, orphaned snapshots, unassociated public IPs, idle load balancers
  • Stopped-but-billed compute; deallocated vs stopped semantics understood
  • Empty or idle clusters, over-provisioned node pools
  • Old snapshots and images beyond retention
  • Dev/test running outside business hours
  • Duplicate or abandoned environments from past projects
  • Log and data retention beyond the actual requirement — often the quiet top-5 line item

Optimize — rightsizing and architecture

  • CPU/memory utilisation percentiles over 30+ days, not point-in-time
  • Instance family modernisation (newer generation, better price/performance, ARM where supported)
  • Storage tiering and lifecycle rules; premium storage on cold data
  • Database sizing, serverless/autoscale tiers, read-replica necessity
  • Egress: cross-region and cross-AZ traffic, NAT gateway processing, CDN offload
  • Managed service vs self-hosted total cost, including the operational hours
  • Autoscaling that actually scales down; minimum instance counts reviewed

Optimize — commitments and rates

  • Steady-state baseline identified before committing anything
  • Reserved instances / savings plans / committed use coverage and expiry calendar
  • Commitment utilisation percentage; unused commitment is worse than on-demand
  • Spot/preemptible for fault-tolerant and batch workloads
  • Licensing: hybrid benefit, BYOL, dev/test pricing
  • Enterprise agreement or negotiated discounts applied to the right accounts

Operate — keep it from regressing

  • Budgets with alerts at meaningful thresholds, routed to the owning team not a mailbox
  • Cost gates in CI: estimated delta on infrastructure pull requests
  • Policy blocking expensive SKUs and untagged resources at creation
  • A recurring review with named owners and tracked actions
  • Savings tracked as realised, not projected — most FinOps programmes fail here

Output

## FinOps Review — <estate/workload>

Current run rate: <x>/month   Trend: <y>%   Untagged: <z>%

| # | Action | Type | Annual saving | Effort | Risk |
|---|---|---|---|---|---|
| 1 | | Waste / Rightsize / Commit | | | |

Total identified: <annualised>
  Waste (delete):        <x>
  Rightsizing (change):  <y>
  Commitments (pre-pay): <z>

### Do this month
### Practices missing
### Data I could not get

Never recommend a commitment purchase before waste and rightsizing — you would be locking in the current inefficiency for one to three years.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.