AI Cost Visibility & Optimization Understand, allocate & reduce your AI costs - Learn More

Introducing the nOps MCP & Skills

Today, we are announcing the nOps MCP.

With the introduction of the nOps MCP, the data layer is no longer limited to the nOps UI. Teams can now connect nOps to MCP-compatible harnesses such as Claude (Desktop, Cowork, and Claude Code), Cursor, Gemini CLI, Amazon Quick, and any other client that supports remote MCP servers.

Why cost questions get stuck outside nOps

  1. The people who change the bill aren't always in nOps. Cloud cost gets decided in a Terraform file, a pull request or an AI coding session, not in a FinOps dashboard. In our own account, a short burst of c8g.16xlarge instances in us-west-2 pushed two days of EC2 spend to $3.6K against $1.7K expected. That's the kind of change an engineer makes from an editor, and the kind of number they only see if they stop and open nOps.
  2. General-purpose AI doesn’t know how nOps does the math. “What’s our savings rate?” has one answer in nOps: Discounted Effective Savings Rate, a defined metric with set rules for how commitments and discounts count. Point a general-purpose assistant at raw billing data and it has to invent that formula on the spot, and it rarely invents the same one twice. You get a number that sounds right, shifts from one question to the next, and doesn’t match the one on your nOps dashboard.
  3. Answers don't turn into anything. Even a correct answer ends in the chat window. Learning that spend is $11.9K above baseline doesn't create the budget, alert or weekly summary that would catch it next time. Someone still has to open nOps and set it up by hand.

One governed connection from your AI tools to the whole nOps platform

The nOps platform, as tools: MCP (Model Context Protocol) is an open standard that lets AI applications connect to outside tools and data. The Clara MCP Server exposes the same building blocks as nOps: Explorer, reports, budgets, cost allocation, recommendations, anomalies, contracts, MAP projects, unit economics, AI spend and scheduled prompts. An assistant can look up the numbers and then act on them, with Clara managing the datasets, permissions and query behavior underneath.

Built for agents, not just chat: Instead of raw SQL, each tool matches how FinOps work gets done in nOps, such as creating a budget from a saved report or listing anomalies for a time window. Every tool says what it does, what it needs and whether it is read-only, so assistants pick the right one. Discovery tools look up valid accounts, services and other values instead of guessing, and analysis runs on LLM-ready functions such as Discounted Effective Savings Rate that return deterministic results.

Skills for the workflows that matter: The Claude and Cursor plugins ship with skills such as /commitment-analysis and /commitment-recommendations that guide the assistant through multi-step Commitment Management workflows and render the results as charts (artifacts in Claude, canvases in Cursor).

Same security as the web app: Every request signs in with your nOps account (OAuth) and goes through the same authentication path as the nOps web application. Every tool is scoped to your nOps organization, and if you belong to more than one, you choose which one Clara can access.

Ask a FinOps question, get an answer you can act on

Here's what that looks like in practice: Claude Desktop connected to the Clara MCP Server, answering four everyday FinOps questions. Each answer comes back from a single plain-English prompt. The labels at the top of each one show which Clara tools did the work, so the numbers come from the same governed data as your nOps dashboard, not from a query the AI wrote itself.

See where your AI spend is going

Start with the question every team is asking right now: how much are we spending on AI, and on what? Claude calls Clara's AI model-usage summary and trend tools and draws the answer.

You get total AI spend, tokens, model count and platforms in one view, with daily usage broken out by model. Bedrock, Claude.ai, Cursor, GCP and other AI vendors show up side by side, so it's easy to see which model is driving the bill and when its usage took off.

Catch the spikes

Next, ask whether anything unusual happened. Claude pulls Clara's detected anomalies and compares actual with expected spend for each one.

Every spike is listed with how far it ran above expected and for how long, ranked so the biggest one is first. Each anomaly traces back to the service, region and usage type behind it, so you know where to look before you open a single console.

Rank the savings

Then the question that matters most: where can we save money? Claude pulls your open nOps recommendations and ranks them by monthly savings.

The answer is a ranked to-do list with the dollar value of each fix, plus a separate view of savings that target AI spend itself, from right-sizing model tiers to trimming long-context prompts and turning on prompt caching.

Explain the bill

Finally, ask for the big picture: the total cloud bill for the last six months, and anything that stands out. Claude queries Explorer history and calls out the outlier.

Clara doesn't just chart the trend, it explains it: a one-time annual commitment that inflates a single month is separated from the services that are steadily growing, so a jump on the bill doesn't get mistaken for a runaway cost.

The bottom line: Four questions, four answers with charts, without opening nOps or writing a query. Every number comes from the same governed data as the nOps dashboard, and the next step, whether that's a budget, an alert or a scheduled weekly summary, is one more request away.

Connect your AI tools to nOps in five steps

1. If you're not on nOps yet, sign up at clara.nops.io.

2. Go to Settings and scroll to Clara MCP & Connectors at the bottom of the page.

3. Click Download assets to get the plugin for Claude or Cursor, or click Copy JSON to add Clara as a standalone MCP server in any other client.

4. Sign in with your nOps account when your client asks. If you belong to more than one organization, pick the one you want Clara to access.

5. Ask away. Try “Run a /commitment-analysis!” or “Show me my recommendations for Commitment Management!”Everything you need is on the Settings -> Plugins page: install guides for Cursor and Claude, the standalone MCP config, and the latest plugin release.

Once you're connected, just ask. Here's Claude Cowork with the nOps MCP plugin installed, ready to run a commitment analysis.

Frequently Asked Questions

Which AI tools can I connect?

Any harness that supports remote MCP servers, including Claude (Desktop, Cowork and Claude Code), Cursor, Gemini CLI and Amazon Quick. Claude and Cursor also get a plugin with skills built in; everything else connects with the standalone MCP config from the Settings page.

What can the assistant do with Clara?

Clara exposes purpose-built tools for the core nOps workflows:

Area

What the assistant can do

Explore spend

Explorer summaries, trends over time, heatmaps, period-over-period comparisons and saved Explorer dashboards

Report and share

Create and update saved reports, render them and send them to stakeholders

Plan and track budgets

Create budgets from a saved report, generate planning targets, compare actuals, add budget alerts and export to CSV

Allocate costs

Build and edit showbacks and their allocation rules, preview an allocation before saving it and review allocation history

Act on insights

Review recommendations and spend anomalies, mark recommendations as implemented, track their results and check an infrastructure-as-code change against open recommendations

Manage commitments and programs

Track contracts and their burndown, and manage AWS Migration Acceleration Program (MAP) projects

Unit economics and AI spend

Business unit economics, AI model usage across platforms and AI seat tracking

Automate

Schedule notifications and recurring Clara prompts

Go deeper with data

Query curated Clara datasets such as Explorer Cost & Usage and Commitment Management Analysis

Which skills come with the plugins?
SkillDescription
/commitment-analysisRun a detailed analysis of your Commitment Management strategy
/commitment-recommendationsGet an overview of potential recommendations for ways nOps can save you more money on your usage by automating commitment management
/query-claraA generalized skill for being able to query your Cost & Usage Data as well as Commitment Management opportunities
Is my data secure?

Yes. Every request goes through the same secure authentication path as the nOps web application, with the same level of data security. You sign in with your nOps account (OAuth), and every tool is scoped to your nOps organization.

How does Clara make sure the numbers are right?

The assistant never writes its own SQL against your bill. It calls tools that match nOps workflows, looks up valid accounts and services instead of guessing, and uses LLM-ready functions such as Discounted Effective Savings Rate that return deterministic results. For deeper questions it queries curated Clara datasets through a structured query of dimensions, measures and filters.

How do I get started?

Navigate to Clara Plugins to install the MCP in your AI harness of choice.

Demo

AI-Powered Cost Management Platform

Discover how much you can save in just 10 minutes!

Book a Demo
Demo
Benjamin White

Benjamin White

Published Date: November 5, 2026, Announcement

Related Posts

Introducing Cursor Integration in nOps

Announcement

Introducing Cursor Integration in nOps

byRick HaggartRick Haggart•Published Date: Jul 13, 2026
Introducing Claude.ai (Enterprise) Integration in nOps

Announcement

Introducing Claude.ai (Enterprise) Integration in nOps

byRick HaggartRick Haggart•Published Date: Jul 17, 2026
Introducing AI ROI in nOps

Announcement

Introducing AI ROI in nOps

byJordan SteinJordan Stein•Published Date: Nov 19, 2026
Introducing Oracle Cloud Infrastructure (OCI) in nOps

Announcement

Introducing Oracle Cloud Infrastructure (OCI) in nOps

byJordan SteinJordan Stein•Published Date: Nov 18, 2026
nOps AI Optimization Recommendations: Closing the Gap From Insight to Verified Savings

AI & Tokenomics

nOps AI Optimization Recommendations: Closing the Gap From Insight to Verified Savings

byRick HaggartRick Haggart•Published Date: Nov 3, 2026
Introducing Databricks Optimization in nOps: Find and fix idle Databricks compute in a few clicks

Databricks

Introducing Databricks Optimization in nOps: Find and fix idle Databricks compute in a few clicks

byJordan SteinJordan Stein•Published Date: Nov 4, 2026