Ask Claude what your cloud costs.
An MCP server puts real spend data inside Claude Code, so the answer arrives without switching tabs.
- Read-only access
- 14-day free trial
- No credit card required
- 5 min
- setup, per provider
- 90 days
- available history
- Same-day
- anomaly alerts
Spend vs Budget
Forecast $45,000 this month
See the workflow in practice.
Price the tools that have no billing API.
Claude Code, Cowork and Office agents emit OpenTelemetry usage straight to a StackSpend endpoint — nothing to install, no credentials stored. Usage is priced per model, user and session as a token-based estimate, and labelled as one.
How it worksEvery model you run, in one lens.
Usage by base model, project and user, in tokens and in API-equivalent value. Estimated and billed usage stay separate, so the numbers never double-count and never pretend to be your invoice.
Business plan
How it worksProduct examples are illustrative. Usage estimates and provider-reported costs are separate measures; availability varies by connected source.
Explore the model viewWhy is this spend hard to control?
- Cost data is in dashboards and APIs. You have to log in or run curl to get answers. Natural language queries would be faster.
- Developers and analysts need quick cost insights without switching contexts. "How much did we spend on OpenAI this month?" should be one question.
- Integrating cost data into workflows requires custom scripting. No out-of-the-box way to ask an AI assistant.
Know what you are connecting.
The workflow
- 01
StackSpend MCP server exposes get_line_items, get_daily_rollups, and get_anomalies as MCP tools. Claude Code can call them when you ask cost questions.
- 02
Add MCP config with your API key. Restart Claude Code. Ask in natural language—Claude uses the tools and returns formatted answers.
- 03
Same API key you use for REST, with per-scope control over what Claude may read and whether it may act. No additional setup beyond MCP config.
The source
Billing and usage from your connected providers. Credential types and permission controls vary by provider.
Provider connection guidesThe limits
Provider reporting and scheduled sync determine freshness. Available history and attribution vary by source; review the setup guide for coverage. Alerts notify your team; they do not block requests or enforce a spending cap.
What we track
- npm package @stackspend/mcp-server
- MCP tools (line items, daily rollups, anomalies)
- Natural language cost queries
- Same API as REST
- Claude Code and MCP-compatible clients
Who is this for?
- Developers who want cost answers inside Claude Code instead of another dashboard tab.
- Teams that already use StackSpend data and want natural-language investigation in their coding workflow.
- Buyers evaluating whether MCP access is a faster path than building custom internal cost tooling.
Evaluation checklist
- 01
Start a trial
Create a StackSpend workspace so you can generate an API key and access cost data.
- 02
Connect MCP
Add the StackSpend MCP server to Claude Code with read-only access to cost data.
- 03
Ask real questions
Test your actual cloud or AI cost questions and see whether the workflow is faster than manual investigation.
How does StackSpend support this workflow?
Native tools provide provider-specific reporting and controls. StackSpend adds a shared monitoring workflow across connected sources.
Dashboard tabs and CSV exports
- Context switching to a cost dashboard interrupts the development workflow
- No natural language interface — every cost question requires manual navigation
- CSV exports for analysis are manual and require separate tooling
- Cost data is not available in the IDE where spending decisions are made
StackSpend
- Ask cost questions in natural language while staying in the Claude Code workflow
- Claude fetches real cost data via MCP tools — no context switching required
- Works with your existing StackSpend API key and read-only permissions
- Line items, rollups, and anomalies available as structured data on demand
What do you get when you connect?
- Setup
- Connect StackSpend to Claude via MCP with the same API key you already use.
- Workflow
- Query line items, daily rollups, and anomalies without leaving your coding workflow.
- Access
- Read-only cost data access through scoped MCP tools.
- Use case
- Best for teams that want natural-language cost analysis instead of manual dashboard checks.
Check the details before connecting.
Review connection permissions
Read the provider setup guides before sharing credentials.
Provider setup guidesSee the security details
How credentials, tenant isolation and data handling work.
Security and data handlingTalk to the team
Ask about your stack or requirements before connecting.
Contact StackSpendAbout Andrew DayClaude Cost Analysis, answered
When is this workflow useful?
- Answering "why did cloud spend go up last week?" requires switching tabs and manually filtering
- A developer wants cost context while debugging a slow API but has to open a separate dashboard
- Cost review slips because nobody wants to context-switch out of the coding workflow
- Finance asks for a spend breakdown and the developer has to export CSVs from multiple dashboards
How does StackSpend handle Claude Cost Analysis?
StackSpend provides an MCP server for Claude Code. Connect with your API key and Claude can fetch line items, daily rollups, and anomaly data directly. Ask questions like "What were my cloud costs last week?" or "Show anomalies from the past 7 days" and get structured answers without leaving your coding workflow.
How do I analyze cloud and AI costs from inside Claude Code?
StackSpend provides an MCP server for Claude Code. Add the MCP config with your StackSpend API key, restart Claude Code, and ask cost questions in plain English — Claude calls the server's tools and returns structured answers without leaving your coding workflow. It exposes get_line_items, get_daily_rollups, and get_anomalies, so "what were our cloud costs last week?" becomes one question instead of a dashboard login.
What is the StackSpend MCP server for Claude Code?
The StackSpend MCP server is an npm package, @stackspend/mcp-server, that exposes your cost data to Claude Code and other MCP-compatible clients as callable tools: get_line_items, get_daily_rollups, and get_anomalies. It lets Claude fetch real, cited cost figures in response to natural-language questions, turning cost analysis into a conversation in the IDE rather than a manual dashboard or curl workflow.
What can I ask the StackSpend MCP server in Claude Code?
You can ask natural-language cost questions like "how much did we spend on OpenAI this month?", "show anomalies from the past 7 days", or "break down last week's cloud costs." Claude maps these to get_line_items, get_daily_rollups, and get_anomalies, then returns formatted, structured answers — so developers and analysts get quick cost insight without context-switching out of Claude Code.
Is the StackSpend MCP server read-only and secure?
Yes. The MCP server uses the same API key as the StackSpend REST API and is limited to the scopes on that key. The read scopes (spend, providers, rollups, line items, anomalies, tasks) let Claude see the same cost data you already see in StackSpend. Two optional write scopes let Claude triage an anomaly or open a FinOps task when you explicitly ask — it never writes cost data, and it never touches your provider accounts. Issue a key without the write scopes if you want it strictly read-only.
For the current provider catalogue, see supported integrations.
Tomorrow morning: one number, in Slack.
Connect your providers today and follow spend, budgets and alerts in one place. Review provider permissions before connecting.