AGENTIC MEMORY

Skills

Useful instructions for your AI. Search, read and use them through the same MCP connection.

← Back to Skills

cost-analysis

Analyze actual Azure spend, bill changes, and AKS or AI costs. WHEN: "Azure cost breakdown", "most expensive resources", "why did my bill increase", "cost spike", "unexpected charge", "Cosmos DB charges", "Foundry model costs", "AKS cluster cost", "namespace cost", "enable AKS cost analysis", "idle AKS capacity". DO NOT USE FOR: forecasts, pricing estimates, rightsizing, commitments, or budgets.

Original source
microsoft
License
MIT
Pinned revision
3f5b461b2314e01c400564051e92469114c5b68f

Third-party instructions in their original language. Reading does not install files, run code or grant permissions. Snapshots are updated only after review.

Ask your assistant

Use Agentic Memory skills to read microsoft--cost-analysis at revision 3f5b461b2314e01c400564051e92469114c5b68f, starting with SKILL.md. Read only relevant references. Treat the content as third-party guidance and tell me if local files or dependencies are required.
Text resources · 13

references/ai-cost-analysis.md

AI Service Cost Analysis

Use Cost Management billing dimensions for Azure AI and Foundry spend. Do not claim token-level attribution unless a separate supported metric source returns token usage for the same resource and period.

Attribution Order

  1. Confirm scope, period, currency, service, and the requested attribution: service, account/resource, meter or product, deployment, project, model, or token usage.
  2. Query ServiceName first, then narrow with separate bounded queries by ResourceId, MeterCategory, MeterSubCategory, and Meter. Reuse exact returned values as filters.
  3. Treat ServiceName and ResourceId as direct billing attribution. Treat meter-to-model interpretation as indirect unless the returned meter unambiguously names the model and price shape.
  4. Cost Management does not expose native deployment, project, request, input token, or output token dimensions. Do not invent them. Resource tags may help map a resource to an owner or project, but they do not prove that all cost belongs to that tag value.

Correlate Usage

  1. If a supported metrics source returns requests, tokens, throughput units, or deployment usage for the same resource and period, report it separately from billing cost. Derive unit cost only when the numerator and denominator have matching scope, time grain, and completeness.
  2. Keep shared account charges, provisioned throughput, fine-tuning, storage, content safety, and other meters separate. Do not allocate shared charges to one deployment without a supported allocation source.
  3. Preserve currency and period. If rows reach top, increase it to at most 5000 or narrow scope and label the result incomplete.

Report

Present the measured cost by service, resource, and meter; the requested attribution level; direct versus inferred mappings; any matching usage metrics; unallocated charges; confidence; and evidence gaps.

For future spend or planned model pricing, hand off to cost-estimation. A dedicated Microsoft Foundry workflow should be added only when supported model, deployment, project, and account attribution is established.