
FinOps agent MCP: one allocated view of cloud, AI and data costs for your whole team, with SSO.
Costory is the FinOps agent your team builds on: a hosted remote MCP for multiple providers and users. - Multiple providers: AWS, GCP, Azure, LLM APIs and data clouds allocated to teams and features in one model, incl. telemetry (Kubernetes, Datadog, dbt). - Multiple users: OAuth, per-user access and enterprise SSO. - Maintained for you: tools, skills and connectors kept up to date by Costory. Finds savings, explains every change in spend and issues recommendations your coding agent can implement. Works in Claude, Cursor, any MCP client and Slack. Customers: Alan, Deezer, Delivery Hero, Voodoo, Pennylane. Pricing: Starter from $300/month, flat, unlimited users and data sources. 30-day free trial.
·
Costory is the cost-data layer for a FinOps agent. It ingests and normalizes cloud, Kubernetes, SaaS, and AI spend (AWS, GCP, Azure, Snowflake, Datadog, OpenAI, Anthropic), allocates it to teams, products, and features, and exposes it through a hosted Model Context Protocol server. Claude, Cursor, VS Code, Codex, Gemini CLI, Dust, or your own agent then investigates, explains, and optimizes spend with structured tool calls instead of raw billing lines.
This repository is the open part: the agent skills and plugin packaging (Claude Code, Codex, Cursor) that teach an agent how to run FinOps workflows on top of the Costory MCP tools.
https://app-api.costory.io/mcp (streamable HTTP)io.costory/finops on the official MCP RegistryA provider billing MCP returns line items for one cloud. That works for a single account. On a real stack (several clouds, shared Kubernetes clusters, many teams) the agent also needs to know who owns which spend, how shared and untagged cost is split, and what changed when the bill moved. Costory maintains that context, so the agent starts from allocated, correlated data.
One click adds https://app-api.costory.io/mcp to VS Code or Cursor; the OAuth login opens in the browser on first use.
Cursor deeplink, if you prefer to paste it: cursor://anysphere.cursor-deeplink/mcp/install?name=costory&config=eyJ1cmwiOiJodHRwczovL2FwcC1hcGkuY29zdG9yeS5pby9tY3AifQ%3D%3D
You need a Costory workspace with billing data connected. A 30-day trial is available.
Claude Desktop / Claude Code / Cursor / VS Code: add a custom connector pointing at https://app-api.costory.io/mcp, then complete the OAuth login in the browser window that opens. Per-client walkthroughs with screenshots are in the MCP docs.
This repo ships an .mcp.json you can copy:
{
"mcpServers": {
"costory": {
"type": "http",
"url": "https://app-api.costory.io/mcp",
"oauth": { "callbackPort": 8080 }
}
}
}Claude Desktop (developer config, claude_desktop_config.json), see alternative JSON config:
{
"mcpServers": {
"costory": {
"url": "https://app-api.costory.io/mcp"
}
}
}VS Code (.vscode/mcp.json):
{
"servers": {
"costory": {
"type": "http",
"url": "https://app-api.costory.io/mcp"
}
}
}Try "Use Costory to show my top cost drivers last month". The agent calls get_context, then query.
If your account has access to several organizations, get_context without a slug returns "Multiple organizations" with the list of slugs. The agent should call list_organizations (or read that list), ask which org to use, and pass that slug to get_context and the following tools. You can also name it upfront: "Use Costory org acme and show my top cost drivers last month".
The server exposes tools in five groups. Names and payloads are versioned; the API reference is the source of truth.
| Group | Tools | What they do |
|---|---|---|
| Orientation | get_context, search, get, suggest_groupby, suggest_usage_metrics, suggest_actions | Discover dimensions, dashboards, metrics, and the right way to slice a question |
| Query | query, list_metrics, list_virtual_dimensions | Cost, usage, metric, formula, and budget queries with period-over-period comparison |
| Allocation | create_virtual_dimension_draft, update_virtual_dimension_draft, preview_virtual_dimension_draft, publish_virtual_dimension, virtual_dimension_overlap_matrix | Define custom cost axes with ordered CEL rules, preview, then publish |
| Reporting | create_report, update_report, preview_report_widget, run_report_now, create_dashboard, update_dashboard | Scheduled Slack, Teams, and email reports plus dashboards built from chat |
| Alerting and events | create_alert, preview_alert, list_alerts, create_event, update_event | Cost and budget alerts, and event annotations for correlation |
Write tools act only inside your Costory workspace. Query scoping follows the calling user's workspace role.
Skills are the workflow layer: each one encodes how to sequence the tools above for a class of question, so the assistant does not have to rediscover it.
MCP skillId | Use when |
|---|---|
bigquery | BigQuery warehouse caveats (on-demand vs slots, physical vs logical, labels) plus the [BigQuery] dashboard template |
cost-change-investigation | Explain a cost change with contribution, timing, usage, metric, event, alert, and terminology evidence |
query | Cost, usage, metric, formula, and budget investigation. Explorer period-over-period only; hands off "what changed" to reports Explain |
virtual-dimensions | Create, edit, preview, and publish custom cost axes with ordered CEL rules |
dashboards | Create or extend dashboards with context-first widget inheritance and overview generation |
reports | Scheduled Slack, Teams, and email reports, and preview-first DIGEST to explain last month's cost |
recipes | Ready-made tracking designs matched to an outcome, then handed off to the skills above to build |
Recipes currently cover BigQuery warehouse routing, GCP spend-based CUDs, budget-vs-actual dashboards, EC2 spike alerts, prod-vs-R&D splits, untagged coverage, marketplace spend, provider credits, namespace cost, compute drilldowns, and period-change explanation. See plugins/costory/skills/recipes/.
This repo also ships a Cursor plugin layout next to the Claude Code marketplace:
.cursor-plugin/marketplace.json — multi-plugin marketplace manifestplugins/costory/.cursor-plugin/plugin.json — Cursor plugin manifestplugins/costory/mcp.json — hosted MCP at https://app-api.costory.io/mcp (OAuth in the client)plugins/costory/assets/logo.png — plugin logoListed on cursor.directory/plugins/costory.
The plugin wires the MCP server and installs the skills in one step.
# Claude Code
claude plugin marketplace add costory-io/costory-finops-mcp-skills
claude plugin install costory@costory
# Codex
codex plugin marketplace add costory-io/costory-finops-mcp-skills
codex plugin add costory@costorySkills only, any agent (skills.sh): installs the seven skills into Claude Code, Cursor, Codex and other agents that read Agent Skills. Connect the MCP separately (see above).
npx skills add costory-io/costory-finops-mcp-skillsKiro: this repo ships a Kiro power (MCP connection + FinOps skill). In Kiro, open the
Powers panel → Add Custom Power → Import power from GitHub and enter
https://github.com/costory-io/costory-finops-mcp-skills/tree/main/powers/costory.
.mcp.json ← ready-to-copy MCP client config
skills.json ← MCP skillId -> SKILL.md path
.claude-plugin/marketplace.json
server.json ← official MCP Registry entry (io.costory/finops)
gemini-extension.json + GEMINI.md ← Gemini CLI extension (MCP connection + context)
powers/costory/ ← Kiro power (plugin.json + mcp.json + skills/costory-finops)
plugins/costory/
.claude-plugin/plugin.json
.mcp.json ← MCP server wired by the Claude Code / Codex plugin
README.md
LICENSE ← Apache-2.0 (copy of the root LICENSE)
skills/
bigquery/SKILL.md
cost-change-investigation/SKILL.md
query/SKILL.md
virtual-dimensions/SKILL.md
dashboards/SKILL.md
reports/SKILL.md
recipes/SKILL.md + recipe libraryget_skill)skills.json maps each MCP skillId to a SKILL.md path. When wiring costory-app, load the file from this repo (or a pinned release), strip optional YAML frontmatter, and return the markdown body.
{
"skillId": "dashboards",
"path": "plugins/costory/skills/dashboards/SKILL.md"
}See AGENTS.md for layout rules, version bumps, and validation. Use SKILL_TEMPLATE.md when adding a skill.
[email protected] · Documentation · Privacy Policy · Terms
Apache-2.0, see LICENSE.
Pick your client and paste the snippet. Each one is the same server, written the way that client expects it.
claude mcp add costory --transport http https://app-api.costory.io/mcp{
"mcpServers": {
"costory": {
"url": "https://app-api.costory.io/mcp",
"type": "http"
}
}
}code --add-mcp '{"name":"costory","url":"https://app-api.costory.io/mcp","type":"http"}'Runs as a hosted service. Your client connects over the network and nothing is installed locally.
Paste this prompt into your agent. It reads this page and does the setup for you.
Read https://aiagentslisting.com/mcp/costory-finops-mcp-skills to learn what the "Costory Finops Mcp Skills" MCP server does and how to install it. Add it to my coding agent's MCP configuration as documented on that page, then confirm the server connects and list the tools it exposes.Agents can also browse this directory over MCP at https://aiagentslisting.com/api/mcp
npx skills add costory-io/costory-finops-mcp-skillsThis server runs locally, so we can't read its tool list over the web yet.
Paste this line near the top of your repository's README. The badge always renders the current score.
[](https://aiagentslisting.com/mcp/costory-finops-mcp-skills)Evidence-first AI URL intelligence agent for crawling, entity resolution, social discovery, SEO, tec
WebScore is a free website scanner & checker — scan any site for SEO, performance, security and acce
v0.12.5v0.12.5: free trial is now 30 daysRelease notesv0.12.4v0.12.4: direct remote MCP config (no mcp-remote)Release notesv0.12.3v0.12.3: one-click install badges, skills.sh, Kiro powerRelease notesv0.12.2Release notesv0.12.1Release notesv0.12.0v0.12.0: plugin ships the MCP server, Gemini CLI extensionRelease notesConnect to the AI Agents Listing MCP server at https://aiagentslisting.com/api/mcp using the HTTP transport and complete the sign-in when it asks. Then call the submit_review tool for the listing "costory-finops-mcp-skills" with a rating from 1 to 5 and a short note on what worked and what did not.No reviews yet
Be the first to share your experience with this tool.
Evidence-first AI URL intelligence agent for crawling, entity resolution, social discovery, SEO, tec
Browser & Scraping
A Model Context Protocol (MCP) server for Grafana.
Monitoring & Observability
WebScore is a free website scanner & checker — scan any site for SEO, performance, security and acce
AI & ML Services
One email a week. New agents, MCP servers and skills, and what is actually getting traction.