Cut cloud bills, allocate costs to teams, plan commitments, build chargeback, detect anomalies, forecast spend, and find waste — across AWS, Azure, and GCP. AI agent skills backed by the FinOps Foundation framework, the FOCUS billing spec, and waste sensor KPIs. Vendor-neutral, multi-cloud. Installable in one command via skills.sh.
npx skills add suan-digital/skillsOne command installs all 9 FinOps skills below. Your agent then routes to the right skill
based on the question you ask ("Cut my cloud bill" → cloud-cost-optimization, "Forecast
next quarter" → cloud-forecasting, etc.).
Works with Claude Code, Cursor, Codex, OpenCode, and 40+ other agents. See INSTALLATION.md for all methods.
All installed by the single command above. Your agent picks the right one based on your question.
| Skill | Use when… |
|---|---|
cloud-cost-optimization |
"Our cloud bill is too high." Right-size compute, optimize storage, tighten utilization. |
cloud-waste-detection |
"Find waste." Idle / unused / oversized resources via standardized waste sensors. |
commitment-strategy |
"Should we buy Savings Plans / RIs / CUDs?" Sizing, term, coverage, exit risk. |
| Skill | Use when… |
|---|---|
cost-allocation-tagging |
"We can't attribute costs to teams." Tag strategy, untagged-spend recovery, K8s labels. |
chargeback-showback |
"Build a chargeback model." Internal billing, cost transparency, shared-cost rules. |
unit-economics |
"Cost per customer / transaction / feature." Cloud COGS, per-tenant cost. |
| Skill | Use when… |
|---|---|
cloud-forecasting |
"Forecast next quarter's spend." Budget vs. actual, capacity planning, scenarios. |
cost-anomaly-detection |
"We had a cost spike." Triage, RCA, alert tuning, guardrails. |
| Skill | Use when… |
|---|---|
cloud-finops |
"Assess our FinOps maturity." All 18 capabilities + framework, FOCUS spec, KPIs. |
Each skill is grounded in the same FinOps Foundation framework material — atomic skills focus on the routing relevant to their problem; the umbrella cloud-finops covers everything for full assessments.
| Repo | Content | License |
|---|---|---|
| finopsfoundation/framework | Capabilities, personas, playbooks | CC BY 4.0 |
| FOCUS_Spec | Billing data specification | Community Specification License 1.0 |
| finopsfoundation/kpis | KPI definitions, waste sensors | CC BY-SA 4.0 |
- "Assess our FinOps maturity" — Walks through all 18 capabilities using Crawl/Walk/Run criteria
- "Our cloud bill is too high" — Routes to utilization-efficiency + cost-allocation capabilities
- "How should I structure Savings Plans?" — Routes to manage-commitment-based-discounts capability
- "We can't attribute costs to teams" — Routes to cost-allocation + manage-shared-cloud-costs
- "We had a cost spike last week" — Routes to manage-anomalies capability
- "Need to forecast next quarter" — Routes to forecasting + budget-management capabilities
- "Multi-cloud cost comparison" — Routes to FOCUS spec + data-normalization capability
- "Identify waste in our environment" — Routes to waste sensors + utilization-efficiency
cloud-finops/
├── skills/
│ ├── cloud-finops/ ← Umbrella full-assessment skill
│ ├── cloud-cost-optimization/ ← Atomic skills (each self-contained)
│ ├── cost-allocation-tagging/
│ ├── commitment-strategy/
│ ├── cloud-forecasting/
│ ├── cost-anomaly-detection/
│ ├── unit-economics/
│ ├── chargeback-showback/
│ └── cloud-waste-detection/
│
│ Each skill directory contains:
│ ├── SKILL.md ← Prompt + routing (only custom file)
│ └── references/ ← Generated from .upstream/ submodules
│ ├── capabilities/ ← 18 capability files
│ ├── personas.md ← 8 FinOps personas
│ ├── focus/ ← FOCUS spec overview, columns, features
│ ├── kpis/ ← KPI defs, waste sensors, reducing-waste, container labels
│ └── playbooks/ ← 6 implementation playbooks
│
├── .upstream/ ← Git submodules (source of truth)
│ ├── framework/ ← finopsfoundation/framework
│ ├── focus-spec/ ← FOCUS_Spec (pinned to v1.3)
│ └── kpis/ ← finopsfoundation/kpis
├── scripts/
│ ├── transform-upstream.py ← Reads .upstream/, writes all skills' references/
│ └── check-submodule-freshness.py ← Submodule drift detection
├── .claude-plugin/marketplace.json ← skills.sh marketplace config
├── INSTALLATION.md
├── CHANGELOG.md
└── LICENSE.md ← CC BY-SA 4.0
Two scripts keep all skills aligned with upstream sources:
transform-upstream.py— Reads.upstream/submodules and writes references into every skill directory. Use--checkto verify committed files match.check-submodule-freshness.py— Compares pinned submodule commits against remote HEAD to detect upstream drift.
See .github/CONTRIBUTING.md. Contributions welcome — especially:
- Updated transcriptions when upstream repos change
- New atomic skills covering additional business problems
- Bug fixes in monitoring scripts
CC BY-SA 4.0 — See LICENSE.md