Summary

CoreStack 2606-v6.5 launches CoreStack TokenOps, FinOps for AI, as its own experience. You can now go beyond high-level AI spend and see which agents and users drive it, across cloud AI services, frontier model APIs, AI gateways, SaaS copilots, coding assistants and agent frameworks. You can see who owns that spend, whether it is efficient, whether it is worth it, and whether it is under control.

This release adds Deep Discovery for Amazon Bedrock and Agentcore, Microsoft Foundry including Foundry agents and the Anthropic spend on Amazon Bedrock and Anthropic Platform. Spend across these AI services, and the agents and org users behind it, is reconciled to the bill and flows into the FinOps framework you already use for cloud: Dimensions, budgets, showback and chargeback, and anomaly detection.

New AI Services widgets bring AI spend trends, forecasts and dimension breakdowns into the same dashboards you use for the rest of your cloud estate. New MCP tools for AI Services let any AI agent, such as Claude or ChatGPT, answer AI spend questions from live platform data, including cost, utilization and inventory, along with Claude-specific prompt analytics. Agents can now look beyond the previous two-month window, and they can answer questions about reservations and savings plans. A rebuilt foundation for the AI Agent also makes the Cloud Commitment (formerly Rate Optimization) agents faster and more consistent.

Core FinOps workflows get deeper and faster. You can review any budget as it was configured in a past month. You can also sort Billing Inventory and the Cost Variation Trend widget by Variance % to surface the biggest cost movers immediately.

The platform is also easier to navigate. The left navigation is reorganized into four sections: Governance, Products, Agent Hub, and Platform. The Agent Hub replaces AI Agents (Beta). The CloudOps Inventory page moves to CoreStack’s current UI framework, with accurate group counts and no cap on total resources. Separately, Resource Posture and Locks in CloudOps are retired in favor of native cloud controls, Inventory, and Tag Governance.

Release Highlights

Standalone TokenOps experience with dedicated dashboards and navigation
Deep Discovery for Microsoft Foundry and the Anthropic Platform API; ENHANCED for Amazon Bedrock
MCP tools for AI Services cost, utilization and logs, plus reservations and savings plans, so any AI agent such as Claude or ChatGPT can query live data beyond the previous two-month window
Faster, more consistent AI Agent with playbook-driven analyses and validated queries
Month-by-month Budget history and one-click Variance % sorting in FinOps
Reorganized left navigation with the new Agent Hub

FinOps

TokenOps Experience

Description: TokenOps now has its own experience, with dedicated dashboards, a TokenOps entry under FinOps in the left navigation, and an onboarding path built for AI & Agent spend. You can use TokenOps on its own, or alongside FinOps for Cloud, SaaS and Kubernetes & AI services. If TokenOps is not included in your product bundle, it appears locked in the left navigation, and your provider admin can choose to unlock it.

Key Capabilities:

Open TokenOps directly from the FinOps section of the left navigation
Dedicated TokenOps dashboards for AI, Agents & users cost, usage and business context attribution
Cloud-native AI services (Amazon Bedrock, Microsoft Foundry) onboard through your existing cloud account onboarding with additional permissions and settings; other AI services onboard through a separate AI services onboarding flow.

Key Benefits:

AI-focused teams get one place to start, without having to set up full FinOps first
FinOps teams see AI Service, Agent & User spend next to cloud spend, in the same platform

Supported Providers: All supported AI services

Deep Discovery for Amazon Bedrock and Microsoft Foundry

Description: TokenOps now runs Deep Discovery for Microsoft Foundry and extends Claude code spend through Amazon Bedrock. It brings together cost, usage, inventory, metrics, logs and events for each service and reconciles spend by AI service, agent and org user to your cloud bill. Unallocated spend shows up explicitly, so you know what is still unattributed.

Key Capabilities:

Spend by AI service, agent and org user reconciled to the AWS and Azure bill, with unallocated spend surfaced
Attribution to business units, applications and agents through CoreStack Dimensions
Budgets, anomaly detection, showback and chargeback, and BillOps charge allocation across AI services, agents and org users
Deeper attribution in Foundry when requests carry agent name, agent ID and tags through Azure API Management
Enabled through your existing AWS and Azure account onboarding, with additional permissions and settings for Deep Discovery
Deep Discovery for Anthropic purchased through AWS Marketplace, with costs attributed down to the user level

Key Benefits:

Spend in Bedrock and Foundry, down to agents and org users, gets the same governance as the rest of your cloud estate
Finance teams can charge AI back to the teams that use it

Supported Providers: Amazon Bedrock (including Bedrock AgentCore), Microsoft Foundry (including Foundry Agents)

Budget Historical Time-Period Filtering

Description: The Budget page now includes a Year/Month filter that lets you review how any budget performed in a prior period without leaving the page, exporting data, or raising a support ticket. Historical views are reconstructed from the budget configuration that was in effect at the time, so mid-month threshold changes are reflected accurately. The filter defaults to the current month and complements the existing Budget History tab.

Key Capabilities:

Select any past year and month to view that period's budget amount, threshold configuration, and actual spend
See budget violations for the selected period, including breach count and severity, with a link to Budget Alerts for details
Use the filter across Tenant, Cloud, Dimension (including Kubernetes filters), and Custom budget scopes
View historical spend in the same cost metric the budget was configured with — Billed, True, or Amortized
Export the filtered historical view as CSV or PDF and compare status month over month with a trend indicator

Key Benefits:

Makes month-end variance reporting and chargeback reconciliation faster by putting prior-period budget conditions one click away
Provides audit-ready history of the exact threshold in force when a violation occurred, supporting compliance reviews
Removes the need for export requests or support tickets to answer "what happened to our budget last month"

Supported Providers: All supported cloud provider.

Sort by Cost Variance in Billing Inventory and Cost Dashboards

Description: You can now sort the Cost Variation Trend widget and the Billing Inventory table by Variance % with a single click, instantly surfacing the highest or lowest cost movers. Sorting is available on every FinOps dashboard type, including multi-cloud, single-cloud, FOCUS (FinOps Open Cost and Usage Specification), and Kubernetes dashboards.

Key Capabilities:

Sort ascending or descending by clicking any Variance % column header
Keep the sort order across all Billing Inventory pages so no high-variance rows are missed
Switch between multiple Variance % columns with exactly one active sort at a time
Rows with no variance data always appear at the bottom, regardless of sort direction

Key Benefits:

Reduces weekly cost review time by removing the need to scan long, unsorted tables
Removes the Excel export workaround during month-end reporting
Lowers the risk of missing cost spikes by keeping the highest-variance items at the top

Supported Providers: AWS, Azure, GCP, and OCI

New AI Services Widgets

Description: The platform cost trend widgets now include AI services, so spend on AI services, agents and org users sits next to the rest of your platform costs. A new cost-by-dimension widget breaks that spend down by business dimension, agent and user, and more chart types are available for AI data. It also supports usage and rate.

Key Capabilities:

AI services cost trend next to platform spend
AI services cost broken down by business dimension, agent and user
More chart types for AI services data like heatmap
Usage and rate as individual widgets or combine them to see together.

Key Benefits:

Spend by AI service, agent and user shows up in the same dashboards you already use for the rest of your cloud estate
Usage and rate next to cost show why spend moved: more usage, or a higher rate, such as a shift to a pricier model
Breakdowns show which business units, agents and users drive AI spend

Supported Providers: All supported AI services

AI Agents

MCP Enhancements for AI Services

Description: AI agents that connect to CoreStack's governed MCP tools can now answer AI spend questions from live platform data. New tools cover AI Services cost and utilization, and agents can now look beyond the previous two-month data window. Reservations and savings plans are also available to connected agents, bringing commitments into the same conversation as usage.

Cost, Inventory, Utilization for both Bedrock, Anthropic
Bedrock service Prompt Analytics
Utilization for cloud services

Key Capabilities:

Query AI Services cost, inventory and utilization for Bedrock Agents and Anthropic API from any AI agent connected to CoreStack's MCP server through new MCP tools.
Prompt analytics: Amazon Bedrock only. Anthropic out-of-scope as of now
- Prompt analytics covers prompts CoreStack ingests from Amazon Bedrock (including Claude models on Bedrock). Prompts sent directly to the Anthropic API are out-of-scope for this release.
In addition to supporting utilization for bedrock, anthropic, we are also supporting utilization for cloud services.

Key Benefits:

Delivers answers based on current platform data rather than static extracts
Reveals longer-term AI spend trends for more reliable analysis

Cloud Commitment Agent (formerly Rate Optimization Agent)

Description: The Rate Optimization Agent in the Agent Hub is now the Cloud Commitment Agent. Only the name has changed, and you’ll see the new name in the left navigation and wherever the agent is referenced.

Key Benefits:

The agent’s name now reflects its focus on cloud commitments

Graphion Agent (formerly AppSecOps Agent)

Description: The AppSecOps Agent in the Agent Hub is now the Graphion Agent. Only the name has changed, and you’ll see the new name in the left navigation and wherever the agent is referenced.

Key Benefits:

The agent’s name now matches the Graphion product it supports

Platform

Redesigned Left Navigation

Description: The CoreStack left navigation has been reorganized into four sections: Governance, Products, Agent Hub, and Platform. Products such as FinOps and Graphion now sit in a dedicated Products sections. CloudOps and Assessments have moved into the Platform section, and Settings, which previously had its own separate section, now sits in Platform as well.

Section	Items
Governance	Cloud Account, Container, AI Services, Data Services, Other Services, Tool Accounts, Guardrails
Products: FinOps	Dashboard, Budget, Billing Inventory, BillOps, TokenOps, Cost Anomaly, Cost Optimizer, Reports
Products: Graphion	Cloud Security, Attack Path Analysis, Compliance, Identity & Access, Applications
Agent Hub	FinOps Agent, Graphion Agent, Assessment Agent, Cloud Commitment Agent, Amazon Bedrock Agent, Anthropic API Agent
Platform	CloudOps, Assessments, Analytics, Settings, Help

Key Capabilities:

Navigate through four clearly labeled sections: Governance, Products, Agent Hub, and Platform
Access agentic capabilities from the new Agent Hub, which replaces AI Agents (Beta) and includes Agents, Agent Outcomes, Agentic Feed, Agent Builder, and Customize
Use a simplified Governance section with Guardrails as a peer item and a new Overview landing page
See TokenOps with a lock icon if it is outside your current bundle; provider admins can unlock it by changing the tenant’s bundle, with no redeploy required

Key Benefits:

Makes features faster to find with fewer menu levels

Angular Migration for Inventory (Phase 1)

Description: The Inventory page (CloudOps → Cloud Inventory → Inventory) has been migrated to Angular, moving it onto CoreStack’s current UI framework. The migration also introduces new capabilities: query-based filtering, a consolidated actions view, and summary sections. Resource listing, filtering, and counts run on a new /v2/inventory/ API set. The summary sections, which feed the group-by and the Untagged tile, are computed over the same results shown in the grid, so group headers show true counts at the provider, category, resource type, and resource levels. The total resource count is no longer capped.

Key Capabilities:

New: Query-based filtering and group-by with accurate counts by provider, category, resource type, and resource, plus a total match count without the previous cap
New: Consolidated actions view for resource actions, utilization, and relationships
New: Summary sections computed over the same results shown in the grid, feeding the group-by and the Untagged tile
Tag management, including the restored Untagged tile with its meter and click-to-filter toggle

Key Benefits:

Brings Inventory onto CoreStack’s current UI framework, consistent with other modernized pages
Helps you find resources faster with the new query-based filtering and act on them from one consolidated actions view
Shows accurate group counts and an Untagged tile that always agrees with the results beneath it
Removes the earlier cap on total resource counts

Supported Providers: All supported cloud providers

Angular Migration for Reports

Description: The Reports page has been rebuilt on CoreStack’s modern user interface through Angular now. One page now holds all your reports, with tabs by category, a Pinned tab for favorites, a Schedules tab, and a built-in search box. Power BI and SSRS reports open right inside CoreStack and load faster and more reliably. You can also subscribe to an SSRS report straight from the report viewer, using the filters you’ve already chosen. Links to the old Reports pages now take you to the new page automatically, so bookmarks and saved links keep working.

Key Capabilities

Reports are grouped into tabs (All, by category, Pinned and Schedules) based on what you’re allowed to see, with search built into the tab bar. You can link straight to a tab.
Pin or unpin any report from the list to keep the ones you use most in the Pinned tab.
Power BI reports and dashboards open inside CoreStack and are filtered to your tenant automatically. SSRS reports open in the built-in report viewer with your report settings already filled in.
If you have permission, a Subscribe button in the SSRS viewer opens the schedule dialog with your current report settings already filled in.
Create, edit and delete report schedules (with a confirmation step before deleting). You can set how often a report runs, who receives it and its file format. Email addresses are checked and duplicate schedule names are blocked.

Key Benefits

Moving from page to page doesn’t reload the whole page, reports load on demand, and refreshing keeps you where you were.
Fewer steps to the right report: Search, category tabs and pinned favorites replace scrolling through a long list.

Midnight Graphite Theme (Dark Theme)

Description: CoreStack now offers Midnight Graphite, a refined, minimal dark theme. You can turn it on from Profile Settings to give the whole CoreStack interface a low-glare graphite look that’s easier on the eyes during long sessions.

Key Capabilities

Adds Midnight Graphite (Refined Minimal Dark) as a theme option in Profile Settings
Applies a consistent dark graphite palette across the CoreStack interface
Saves the theme to your profile, so it stays applied each time you sign in

Key Benefits

Cuts eye strain in low-light settings and during long work sessions
Lets you personalize the workspace without affecting other users

Deprecated Items

Resource Posture

The Resource Posture capability (the Resource Governance dashboard) in CloudOps was retired, along with its List Resource Posture Details public API endpoint. Because Resource Posture was read-only, no customer configuration was lost. Resource counts remain available in Resource > Inventory, and naming-standards compliance is available in Tag Governance.

External APIs

To see the external APIs which have been added, modified, and removed in this release, refer to: External APIs 6.5 (2606)
To see all the available external APIs, refer to: https://docs.corestack.io/reference/accesssummarycount