MCP App Store
by KeenEthics
Infracost icon

Infracost MCP Server for Claude and ChatGPT

By Infracost

Overview

Connect your Infracost organization to bring your FinOps practice into every conversation. Check which FinOps and tagging policies and cost guardrails apply before you change infrastructure. Investigate cost anomalies and waste Infracost has found in your cloud, including which pull requests caused them and how to fix them. Uses your existing Infracost account and permissions.

How to connect

  1. Claude: Open the Infracost listing in the Claude connectors directory and choose Connect. Claude listing Infracost exposes 17 tools here.
  2. ChatGPT: Open the Infracost listing in the ChatGPT app directory and choose Connect. ChatGPT listing Infracost exposes 17 tools here.

Tools

create_fix

ChatGPT
Create a PR or ticket from a reviewed preview_fix configuration. org_id is required. This has external side effects and requires explicit user confirmation.

get_finding

ChatGPT
Fetch a single FinOps finding by id for the required org_id. Returns the finding header plus its nested tasks, actions, and events — including each task's action description and code snippet. Use preview_fix to draft a task fix, then call create_fix only after user approval.

get_guardrail

ChatGPT
Get one cost guardrail by ID for an organization, including its scope, filters, thresholds, non-secret actions, and trigger summary.

get_issue

ChatGPT
Get one Dashboard policy issue by ID for an organization, including policy and repository context, resource evidence, savings, source location, current status, and lifecycle events. This does not return an Agents finding or remediation tasks.

list_cloud_security_policies

ChatGPT
List cloud security policies configured for an organization. Policies include effective metadata, applicability filters, custom settings, PR actions, and risk, effort, and downtime guidance.

list_findings

ChatGPT
List FinOps findings for an organization by view (default: active work queue), not exact status. The resolved view includes resolved, dismissed, and effectively suppressed findings; dismissed includes only explicit dismissals. There is no all-findings or duplicate view. Totals cover the selected view and filters, not the entire organization. org_id is required. Each row includes its ID, title, summary, status, estimated yearly savings, and task counts. Use page to continue from next_page, keeping the same view and effort.

list_finops_policies

ChatGPT
List FinOps policies configured for an organization. Policies include effective metadata, applicability filters, custom settings, PR actions, and risk, effort, and downtime guidance.

list_guardrail_events

ChatGPT
List guardrail trigger events for an organization, newest first. Optionally restrict the result to one guardrail. Events include the triggering reason, thresholds captured at trigger time, projects, configured actions, block state, related guardrail and run IDs, and source repository, pull-request, and exact-commit provenance when available. Use the provenance with VCS access to inspect the triggering change; it does not authoritatively attribute individual resources.

list_guardrails

ChatGPT
List configured cost guardrails for an organization. Each guardrail includes its scope, filters, thresholds, non-secret actions, and trigger summary. Use get_guardrail for one complete guardrail.

list_issues

ChatGPT
List Dashboard policy issues for an organization. These are repository/resource policy violations, not Agents findings. Defaults to currently open issues and supports policy type and lifecycle filters. Use get_issue for detailed evidence and history.

list_organizations

ChatGPT
List the organizations available to the authenticated user, including each organization's ID, name, and slug. Use an ID from this response as org_id for organization-scoped tools, and the matching name when talking to the user. The preferred_organization_id identifies the user's default.

list_tagging_policies

ChatGPT
List tagging policies configured for an organization. Policies include their scope filters, PR actions, compliance setting, and normalized tag requirements.

preview_fix

ChatGPT
Draft a PR or ticket for a task without creating it. org_id is required. Show the returned type and config to the user before passing them to create_fix.

retry_action

ChatGPT
Re-queue a failed PR or ticket action for the required org_id.

update_finding_status

ChatGPT
Set a finding's status to open, resolved, or dismissed for the required org_id. A dismissal reason also teaches the agent not to re-raise the same finding.

update_task_status

ChatGPT
Update a task's resolution for the required org_id. status='confirm' says "I did Agents' suggested change" — call it once the change is made. It marks the task done, the same as Mark as done in the app, and the next scan verifies it; no create_fix call is needed first. status='correct' says "I solved the problem a different way" — dismisses the task with reason as the dismissed_reason, cascade-dismisses any draft/open actions linked to it, and emits an AgentLearning(source=correction). status='dismiss' says "we're not going to do this task" — dismisses the task with reason as the dismissed_reason and emits an AgentLearning(source=task_dismissed). Reason is required for correct, recommended for dismiss (needed to emit the learning), and ignored for confirm.

whoami

ChatGPT
Return the authenticated subject's ID, name, email address, and preferred organization ID. The subject is a Dashboard user, or an organization service account when the request carries an organization token, in which case the email is empty. Use list_organizations to inspect the organizations available to it.

create_fix

Claude

findings_get

Claude

findings_list

Claude

get_guardrail

Claude

get_issue

Claude

list_cloud_security_policies

Claude

list_finops_policies

Claude

list_guardrail_events

Claude

list_guardrails

Claude

list_issues

Claude

list_organizations

Claude

list_tagging_policies

Claude

preview_fix

Claude

retry_action

Claude

update_finding_status

Claude

update_task_status

Claude

whoami

Claude

Similar Apps

Listed on MCP App Store

Listed on MCP App Store

Add this badge to your site or README. It links to this page.

HTML

<a href="https://mcp-app-store.keenethics-labs.com/app/infracost"><img src="https://mcp-app-store.keenethics-labs.com/badges/listed-on-mcp-app-store.svg" alt="Listed on MCP App Store" width="220" height="44"></a>

Markdown

[![Listed on MCP App Store](https://mcp-app-store.keenethics-labs.com/badges/listed-on-mcp-app-store.svg)](https://mcp-app-store.keenethics-labs.com/app/infracost)