FinOps MCP Servers — Usage Guide

Source on GitHub ↗

Page 2 of 7 · Server reference

finops-framework-mcp

An agentic interface to the official FinOps Framework (finops.org/framework): 6 Principles, 3 Phases, 4 Domains containing 22 Capabilities, 11 Personas, 5 Technology Categories, Scopes, and an 88-entry KPI library — each Capability carrying a definition, Crawl/Walk/Run maturity assessments, per-persona activities, and featured KPIs.

finops-framework-mcp · data v2.1.1, crawled 2026-07-23 11 tools · 4 prompts · 11 fixed + 4 template resources

framework · get_framework_info {}

Data model

The server's own orientation call lays out the hierarchy an agent walks — domains group capabilities, and each capability is the unit that carries KPIs, personas, and maturity guidance:

“4 Domains containing 22 Capabilities (the unit of practice — each with a definition, Crawl/Walk/Run maturity assessments, per-persona activities, and KPIs), 11 Personas (6 core + 5 allied), 5 Technology Categories, the Scopes concept, and a KPI library (88 entries).”
4
Domains
→
22
Capabilities
→
88
KPIs
11
Personas
3
Maturity levels
(Crawl/Walk/Run)
6
Principles
3
Phases
5
Technology categories

The four domains, per get_framework_info: Understand Usage & Cost (4 capabilities), Quantify Business Value (5), Optimize Usage & Cost (5), Manage the FinOps Practice (8). Phases (Inform/Optimize/Operate) are an iterative lifecycle, not a partition of capabilities — the server does not offer a phase filter for that reason.

framework · get_framework_info {} → structuredContent.counts, overview_md

Tool tour

11 tools, grouped by what they're for. Names and parameters below are verified against the live tools/list response (source: docs/mcp-surface.md, re-checked against a live probe for this page).

Discover
ToolPurposeKey params
get_framework_infoOrientation: entity counts, data version/crawl date/license, how to navigate. No params.—
list_capabilitiesAll 22 capabilities (slug, title, domain, summary), optionally filtered.domain, persona, limit (≤50, default 50), cursor
map_personasPersona↔capability matrix. With persona: capabilities that persona works in, with activity bullets. With capability: personas active in it. With neither: full persona index.capability, persona
Read
ToolPurposeKey params
get_capabilityOne capability's content, section-selectable to control size (default [summary, definition] ≈1–2k tokens; all sections ≈7k).capability (required); include: summary|definition|maturity|activities|kpis|headline_groups|inputs_outputs; persona (filters activities)
get_kpisFull KPI records: description, formula + data sources, related capabilities, featured-on list. Single lookup or filtered/paged library browse.slug, capability, featured_only (default false), limit (≤100, default 25), cursor
get_maturity_assessmentOfficial maturity assessment prose for a capability, verbatim, at one level or all three.capability (required); level: crawl|walk|run (omit for all three)
get_maturity_modelThe official Crawl/Walk/Run maturity model itself — capability-agnostic characteristics and sample goals/KPIs per level. No params.—
get_entityFull markdown for entity types with no dedicated tool.entity_type (required): principles|phases|domains|technology-categories|scopes|persona; slug (required only for persona)
get_changelogRolling per-crawl diff summaries of this server's data artifact (crawl-to-crawl, capped at 20 entries).limit (≤20, default 5)
Search
ToolPurposeKey params
search_frameworkRanked keyword search over every entity type. Returns slug + uri per hit — feed slugs into get_capability/get_kpis/etc.query (required, len ≥2); entity_types (filter); limit (≤50, default 10); cursor
Assess
ToolPurposeKey params
assess_maturity_pathOfficial maturity assessment text at each level between a current and target level — evidence to look for when maturing.capability (required); current_level (required, exclusive); target_level (required, must be above current)

Prompts & resources

Prompts (4)

PromptPurposeArguments
explain-frameworkGuided onboarding tour of the framework, tailored to an audience.audience (optional)
assess-capability-maturityStructured interview to place an org's maturity for one capability, citing official assessment text as evidence.capability (required)
plan-maturity-roadmapOrdered plan from a current to a target maturity level, citing official assessment text as evidence.capability, current, target (all required)
map-personas-to-capabilitiesEngagement guide: what a persona (or every persona) does across the framework's capabilities.persona (optional)

framework · list-prompts (live probe, verbatim descriptions/arguments above)

Resources — finops://framework/…

11 fixed resources plus 4 URI templates (44 concrete resources listed when templates are expanded). Fixed resources are whole small documents; templates are per-entity documents completed via slug (and, for maturity, level).

URIWhat it isType
finops://framework/overviewStart here: what the framework contains, how to navigatefixed
finops://framework/principlesThe 6 FinOps Principlesfixed
finops://framework/phasesThe 3 FinOps Phases (iterative lifecycle)fixed
finops://framework/domainsThe 4 Domains with their capabilitiesfixed
finops://framework/technology-categoriesThe 5 Technology Categoriesfixed
finops://framework/maturity-modelOfficial Crawl/Walk/Run maturity levelsfixed
finops://framework/scopesFinOps Scopes — conceptual guidancefixed
finops://framework/personasAll 11 personas (6 core + 5 allied)fixed
finops://framework/capabilitiesAll 22 capabilities grouped by domainfixed
finops://framework/meta/manifestData manifest (JSON): versions, crawl date, counts, source URLs, attributionfixed, JSON
finops://framework/meta/changelogCrawl changelog (JSON)fixed, JSON
finops://framework/capabilities/{slug}Full capability document: definition, maturity, activities, KPIs, inputs/outputstemplate, completable
finops://framework/personas/{slug}One persona's goals, objectives, capability involvementtemplate, completable
finops://framework/capabilities/{slug}/maturity/{level}One capability's official assessment at one leveltemplate, completable
finops://framework/kpis/{slug}Full KPI recordtemplate, completable

framework · list-resources (11 fixed confirmed live, incl. finops://framework/capabilities/anomaly-management) · docs/mcp-surface.md (template/completion detail)

Pagination & cursors

List/search tools (list_capabilities, get_kpis, search_framework) return an opaque cursor string alongside results once more remain; pass it back as the cursor argument to continue. A live 3-per-page walk of the 22 capabilities:

framework · list_capabilities {limit: 3} → "Showing 3 of 22 — pass cursor: "eyJ2IjoiMi4xLjEiLCJvIjozLCJoIjoieUYtbWsxYkNia09wIn0" for more." → structuredContent.nextCursor: "eyJ2IjoiMi4xLjEiLCJvIjozLCJoIjoieUYtbWsxYkNia09wIn0"

The three capabilities returned were allocation, anomaly-management, architecting-workload-placement — pagination is stable ordering over the 22-entry list, cursor opaque and version-scoped (embeds the data version, so a re-crawl invalidates stale cursors rather than silently reordering results underneath them).

Official vs. UNOFFICIAL, and provenance

Out of the box, everything this server returns is official restructured FinOps Foundation content — the Framework, KPI library, personas, maturity model, all as published at finops.org/framework. Every tool response carries a provenance footer citing the source URL, CC BY 4.0 license, and the data version/crawl date, e.g.:

Source: https://www.finops.org/framework/capabilities/anomaly-management/ — © FinOps Foundation, licensed CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). Content restructured and adapted by finops-framework-mcp (data v2.1.1, crawled 2026-07-23); unofficial extensions are always marked.

What a fresh npx finops-framework-mcp serves — every tool, prompt and resource on this page — is content from the official framework pages, restructured and adapted under CC BY 4.0. Nothing this server invented is mixed in, and the content records that make up an answer carry their own source_url in structuredContent alongside the footer above, so any claim can be taken back to the page it came from.

Worked demo: Anomaly Management end to end

The chain an agent walks for one capability: get_capability for what it is, get_kpis for what to measure, get_maturity_assessment for how to grow it.

1

get_capability official

“Detect, identify, alert and manage unexpected or unforecasted technology cost and usage irregularities in a timely manner to lower risk and ensure cost-effective operations.”

The full definition (trimmed here) situates it: “Anomaly Management gives a FinOps team the ability to detect, identify, clarify, alert on, and manage unexpected cost events in a timely manner […] Effective Allocation metadata are critical to effective anomaly detection […] Use of automated, machine learning–based anomaly detection is typical.” Domain: understand-usage-and-cost.

framework · get_capability {capability: "anomaly-management", include: ["summary","definition"]}

2

get_kpis, featured only official

Three KPIs are featured on the capability's own page:

Anomaly-Detected Cost Avoidance featured
Measures amount of cost avoidance as a result of identifying and correcting an anomaly.
Total cost of the unpredicted & correctable amount of spend * Projected Amount of Days the anomaly could have occurred if the underlying issue was not addressed
Data source: Monitoring Tools.
Anomaly Detection Rate featured
Measures the frequency and cost impact of anomalies in AI spending, such as sudden cost spikes or unexpected usage patterns.
Total Cost of Anomaly Spikes / Total AI Spend = Anomaly Cost % where (adjust for your needs): Green (< 2%): Healthy. Normal fluctuations. Yellow (2-7%): Warning. Minor anomaly trend Red (> 7%): Critical. You have a "runaway" costs.
Data sources: API usage reports, dashboards/logs from AI platforms, cloud billing data. Related capabilities: anomaly-management, data-ingestion, reporting-analytics.
Total Unpredicted Variance of Spend featured
Measures the unpredicted variance of effective cost associated with CSP Cloud usage recorded over a given period of time.
Total effective cost associated with all anomalies events detected less the predicted spend of the services related to the identified anomalies
Data source: Monitoring Tools.

framework · get_kpis {capability: "anomaly-management", featured_only: true} → total: 3

3

get_maturity_assessment, all three levels official

Crawl: “Manually checking for anomalous spending using reporting […] Reacting to anomaly activity more than a week after it occurs (post receipt of billing data) […] Anomaly alerts sent to a central team, or require manual action to look.”

Walk: “Some form of automated detection or reporting or tooling […] Context-relevant thresholds are detected (percentage of spend change, single item spend amount ceilings, forecast breach alerts, etc.) […] Unexpected spending automatically routed to responsible teams […] KPIs associated with Anomaly Management set and in use by key teams in the organization.”

Run: “Mature Anomaly detection tooling in use and embedded in tooling across the organization […] Automation created to detect, suggest resolution, or resolve anomalous spending alerts […] Analysis results in full root cause analysis post-mortem where appropriate […] Analysis of resolution of past anomalies improves alerting on future anomalies.”

framework · get_maturity_assessment {capability: "anomaly-management", level: "crawl"} · {level: "walk"} · {level: "run"}

What this demo shows

Three tool calls, zero invented data, one complete practitioner briefing: what Anomaly Management covers, which three KPIs the Foundation features on its page (with exact formulas), and the verbatim Crawl→Walk→Run evidence to look for when maturing the practice — the same chain assess-capability-maturity and plan-maturity-roadmap automate as prompts.

Next: pair this with FOCUS spec data on finops-focus-mcp, or see both servers combined in the Rate Optimization (ESR) worked example.