From ccds-infra
FinOps specialist. Owns cloud cost visibility, allocation, unit economics, right-sizing, commitment strategy (RI/SP/CUD), and forecasting. Auto-invoked for cost investigation, reserved-capacity planning, or unit-economics work.
How this skill is triggered — by the user, by Claude, or both
Slash command
/ccds-infra:infra-finopsThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
Cloud spend is the product of a thousand small decisions. Without allocation,
Cloud spend is the product of a thousand small decisions. Without allocation, attribution, and unit economics, costs drift up silently — and the bill is the first anyone hears of it.
service, team, env, cost-center) via policy (tag policies / SCP
/ org policy), not convention. Target ≥ 95% of spend allocatable; report the
untagged remainder as its own line so it can't hide.| Workload shape | Instrument | Term |
|---|---|---|
| Stable baseline, known family/region | Standard RI / resource-based commitment | 1–3 yr (3 yr only if architecture is settled) |
| Stable spend, changing instance types | Compute Savings Plan / flexible CUD | 1 yr first, extend on evidence |
| Interruptible batch / CI / stateless | Spot / preemptible + fallback to on-demand | n/a |
| Spiky, unpredictable | On-demand; revisit when 3 months of usage data exists | n/a |
Track two numbers monthly: coverage (% of eligible usage under commitment) and utilization (% of commitment actually consumed). Coverage low = paying on-demand tax; utilization low = paying for air. Healthy is both ≥ 80–90%.
Related: infra-sre (reliability-vs-cost trade-offs), infra-k8s (workload
sizing and bin-packing), infra-observability (cardinality and retention cost),
infra-networking (egress paths) · domain agent: infra-architect · output/ADR
format: playbook-conventions
npx claudepluginhub ggrace519/claude-code-dev-studio --plugin ccds-infraProvides behavioral guidelines to reduce common LLM coding mistakes, focusing on simplicity, surgical changes, assumption surfacing, and verifiable success criteria.
Searches, retrieves, and installs Agent Skills from prompts.chat registry using MCP tools like search_skills and get_skill. Activates for finding skills, browsing catalogs, or extending Claude.