FOCUS MCP Server

[helixoft]

Hi folks, I would like to start / submit a discussion for FOCUS (cloud agnostic) specific MCP server that provides specific guidance for framework interaction and customer advisories.

[OptimNow]

Thanks for opening this @helixoft! This echoes the conversation in Discussion #2 . To move this forward, it would help to flesh out the scope a bit:

1. What FOCUS use cases do you see the MCP server covering? (e.g. taxonomy validation, cost allocation mapping, customer assessment)
   2 Would it work against raw billing data or pre-processed FOCUS-formatted datasets?
2. Any thoughts on which FOCUS spec version to target?
   Happy to collaborate on the design. If you want to sketch out a rough proposal or user stories, feel free to drop them here or in the discussion thread.

[helixoft]

1. Having a conversation with customers on Finops is a complex one. Customers want to save money, and want to know "how". To answer this question we need to get through the many layers of: (a) which cloud, (b) which department, (c) which vendors, (d) which projects, and finally (e) which teams. It's a huge amount of org chart mapping, stakeholder analysis, data collection and interaction that can take weeks and months. This is where we want to start to use the Finops FOCUS framework mapped to a proposed engagement lifecycle (LIFECYCLE 1: STAKEHOLDER ENGAGEMENT). Allowing the MCP server to ingest unstructured org and stakeholder data can produce (or populate) a prescribed framework, and identify focused relationships (and gaps) - and do this in a much shorter time. From there, LIFECYCLE 2: APPLYING THE FOCUS FRAMEWORK to generate dimensional data sets: This is assumes, the client is well versed in the FinOps Foundation framework, but start to add their organizational content/data to it (taxonomy, cost center mapping, metrics (number of apps, numer of developers, number of testers, number of end-users, revenue generating index, number of end-customer-served); The key "holy-grail" calculation is how much $ of revenue is generated from $ cost of cloud compute (across the business); and finally LIFECYCLE 3: Using the Finops dimensional data sets to generate dashboards for metrics, decision making and remediation steps.

1.2. Can be both (raw data sets initially for unstructured data sets; and structured data sets that conform to FOCUS-formatted)

2. No preference (but if there are any customer case studies on a particular version).

2.1. I will create a wiki (in HTML form) and start to add it on here (as a repo), and we can collaborate and maintain this wiki. I'm targeting end of April to have a version the community can agree as a draft starting point. A v1 that does not need to be complete (or fully exhaustive) but one that points us in the right direction and is customer friendly (i.e. not too much technical jargon), but answering the how in the "how do we save money". We then break the "how" into the 3 LIFECYCLE phases, and for each phase explain the (a) stakeholders, the data, the engineering required, the MCP servers used, and then populate the dashboards for decision-making and remediation. This becomes a LIVE tracker for progress.

I'm of the oppinion that many (if not or all) of the MCP servers listed on this community can be used at different phases of the lifecyle (or different components / use cases within a phase); and we can then create a mapping (of existing MCP fit), and then derive a roadmap for future MCP extension.

Conclusion:

Customers what a trustworthy business experience through the 3 phases, with clear outcomes. The advisory and remediation must match the promises made at the beginning of the engagement (e.g. we can help save you 20% cloud cost).

Consulting partners must deliver this through mature application of the Finops framework, a well-defined lifecycle (mapped to crawl, walk, run), and "pragmatic" application of MCP servers, agentic, ML APIs to shorten timeframes, become wholy data-centric, and increase the integrity and fidelity of the entire process.

Argument:

I've seen alot of Finops teams deep-dive into cloud billing data, tagging, and recommended savings plans and scheduled server shut-downs, only to be back to square one, a few quarters later due to change and regression. These exercises can be become quite myopic or silo'd - or in the hands of few for decision-making. This is where a more robust organizational-wide platform can add value.

[helixoft]

I have run my proposed lifecycle proposal against the latest Finops framework (using claude) and the alignment as follows checkout. Wiki to follow:

Claude:

I've researched the FinOps Foundation framework thoroughly and have solid context on your vision and GitHub discussion. Here's what I picked up from finops.org:

• The 2026 FinOps Framework has evolved significantly — it now frames FinOps as strategic business partnership rather than just cloud cost management.

• The three operating phases are Inform → Optimize → Operate, with maturity measured as Crawl → Walk → Run across four domains (Understand Usage & Cost, Quantify Business Value, Optimize Usage & Cost, Manage the Practice).

• A major 2026 addition is Executive Strategy Alignment as a formal capability, and FinOps Scopes are now treated as decision contexts tied to business constructs rather than just spending views.

• The framework also now explicitly covers AI, SaaS, data platforms, and data centers — not just public cloud.The FOCUS spec (v1.3) provides the standardized billing schema that normalizes data across providers, with contract commitment tracking, split cost allocation, and dataset completeness verification.

Conclusion
Your 3-lifecycle model maps nicely onto this — your Lifecycle 1 (Stakeholder Engagement) aligns with the Inform phase and the new Executive Strategy Alignment capability, Lifecycle 2 (Framework Application) maps to the dimensional data and FOCUS schema work, and Lifecycle 3 (Dashboards & Remediation) maps to Optimize/Operate.

Caveate
The lifecycle terms (stakeholder engagement, application of framework, remediation) does not replace the finops framework (taxonomy) but is the "action oriented" applied form in a customer context. This is what a consultancy would provide as a service.

[knaicker-dvt]

finops-focus-lifecycle-wiki_2026-04-12_v1.0.html

Good Day, please see v1 of the proposed Wiki, which covers:

1. Summary (the Challenge, Approach and Outcome) - subject to community scrutiny and and consensus
2. Overview (mapping the lifecycle approach to the framework)
3. A breakdown of each Lifecycle component (and use cases)
4. A mapping of all existing MCP servers
5. A proposed Community Roadmap
6. Customer Case Studies (Coming soon!)

To contribute:

1. Discuss with Jean if we can move this to it's own repo
2. Click on the "How to contribute" for a step-by-step guide how to contribute, scrutinize and vett this wiki
3. Update the release notes (see hamburger menu, top right)
4. Request a PR to update (2x approval by Jean + X)

[knaicker-dvt]

TODO:

12-Apr (Kevin Naicker) - Whitepaper - Finops Tokenomics (Model Efficacy vs. token costs vs. business outcome - supported by empirical data)

[knaicker-dvt]

Finops Challenges:

https://www.finops.org/wg/adopting-finops-avoiding-pitfalls/
https://medium.com/@restlessankyyy/why-finops-is-failing-the-enterprise-and-what-should-replace-it-6e0d4c429e68
https://www.linkedin.com/posts/victor-garcia-rubio_your-finops-team-is-failing-i-know-why-activity-7308434837462237188-DQbz/
https://blog.shi.com/next-generation-infrastructure/cloud/finops-management/
https://www.infoworld.com/article/3829451/why-enterprises-fail-at-finops.html