About Chartsy
Chartsy is a read-only analytics MCP server for SaaS businesses, helping agents inspect revenue, churn, customers, and site traffic from connected payment and custom sources. Its companion skills guide concise analytics queries and a weekly, revenue-informed marketing planning loop.
Key features
- 01Read revenue and MRR metrics
- 02Review churn and at-risk customers
- 03Analyze site traffic and sources
- 04Compare revenue by marketing source
- 05Generate business and daily snapshots
- 06Use skills for revenue-led marketing planning
Use cases
- 01Review changes in MRR by plan
- 02Identify customers at risk of churn
- 03Find traffic sources that convert to signups
- 04Draft next week's posts from revenue results
MCP support
Chartsy provides a hosted, read-only MCP server at https://mcp.chartsy.app/mcp using Streamable HTTP and OAuth. Its tools expose analytics for the accounts and sites associated with the signed-in account.
Configuration
Remote servers need no local install. Pick your client, then copy the config or install in one click.
https://mcp.chartsy.app/mcpclaude mcp add --transport http app-chartsy-chartsy https://mcp.chartsy.app/mcp{"mcpServers":{"app-chartsy-chartsy":{"type":"http","url":"https://mcp.chartsy.app/mcp"}}}Open in Cursor{"servers":{"app-chartsy-chartsy":{"type":"http","url":"https://mcp.chartsy.app/mcp"}}}Open in VS Code{"mcpServers":{"app-chartsy-chartsy":{"type":"http","url":"https://mcp.chartsy.app/mcp"}}}{"mcpServers":{"app-chartsy-chartsy":{"serverUrl":"https://mcp.chartsy.app/mcp"}}}Run commands in a terminal; paste JSON into the client’s MCP config and restart. Servers that need sign-in will prompt on first connect.
Tools
Details
From the project READMEChartsy for AI agents
Ask your AI agent about your SaaS business: MRR, churn, revenue, refunds, at-risk customers, goals, site traffic and signups. The numbers are live from Chartsy, which reads Stripe, Paddle, Polar, Lemon Squeezy, Creem, Freemius or a custom source.
This repo bundles what every agent install points at:
- The hosted Chartsy MCP server:
https://mcp.chartsy.app/mcp. It uses streamable HTTP and OAuth sign-in, and it's read-only. Nothing to install or run locally. - The
chartsyskill (skills/chartsy/SKILL.md). It teaches the agent which tool answers which question, how to keep tool calls to a minimum, and how to read the numbers correctly. - The
chartsy-growth-loopskill (skills/chartsy-growth-loop/SKILL.md). It guides your agent through a weekly loop: find which channels, campaigns and posts brought paying customers who stayed, recommend where to focus, and write next week's posts with UTM-tagged links. If you also connect a publishing tool's MCP (e.g. Postiz or Publer) to the same agent, the skill tells it to save the posts there as drafts for your review. Chartsy itself is read-only and never posts anything.
Usage: every tool call counts as one ChartsyAI question against your plan's monthly allowance, the same budget as questions asked inside Chartsy. The skill steers agents to one-call answers (
get_business_snapshot,get_daily_snapshot) wherever possible.
You need: a Chartsy account (any plan, including the free trial) with a payment provider connected and synced.
Install
Claude (claude.ai, desktop, mobile)
Settings → Connectors → Add custom connector, then paste https://mcp.chartsy.app/mcp
and sign in to Chartsy.
Claude Code
Plugin (MCP server and skill):
/plugin marketplace add sprifti/chartsy-agent
/plugin install chartsy@chartsy
Or just the server:
claude mcp add --transport http -s user chartsy https://mcp.chartsy.app/mcp
Then run /mcp to sign in.
Cursor
Add to Cursor.
Or go to Settings → MCP → Add server, choose type HTTP, and use URL
https://mcp.chartsy.app/mcp.
ChatGPT
Settings → Connectors (Developer mode) → add https://mcp.chartsy.app/mcp and sign
in to Chartsy.
Gemini CLI
gemini extensions install https://github.com/sprifti/chartsy-agent
Codex, OpenClaw and other skill-aware agents
npx skills add sprifti/chartsy-agent
The skill needs the MCP server connected in the same agent. Add
https://mcp.chartsy.app/mcp in that agent's MCP settings.
Any other MCP client
Use mcp.json, or point the client at https://mcp.chartsy.app/mcp
(streamable HTTP, OAuth sign-in).
Clients that only run local (stdio) servers
Older clients can't connect to a hosted server directly. Bridge to it with
mcp-remote, which handles the OAuth
sign-in in your browser:
{
"mcpServers": {
"chartsy": {
"command": "npx",
"args": ["-y", "mcp-remote", "https://mcp.chartsy.app/mcp"]
}
}
}
Agents that run without a browser (servers, cron jobs, CI)
OAuth needs someone to click Approve in a browser. For an agent running unattended, such as OpenClaw on a VPS, a scheduled job or a CI pipeline, use a personal token instead:
- In Chartsy, go to Settings → Claude & MCP → Generate token. It's shown once, so copy it then.
- Store it as a secret (e.g.
CHARTSY_MCP_TOKEN). Never commit it. - Send it on every request:
For example with Claude Code:Authorization: Bearer <your-token>
Or throughclaude mcp add --transport http -s user chartsy https://mcp.chartsy.app/mcp \ --header "Authorization: Bearer $CHARTSY_MCP_TOKEN"mcp-remote:npx -y mcp-remote https://mcp.chartsy.app/mcp --header "Authorization: Bearer ${CHARTSY_MCP_TOKEN}"
Tokens last a year and can be revoked on the same tab. An unattended agent uses up ChartsyAI
questions like any other client, one per tool call, so schedule it sensibly. A daily
get_daily_snapshot is one question a day.
Things to ask
- "Give me a full business review. Where's the risk?"
- "What happened in my business yesterday?"
- "How has MRR moved by plan over the last 6 months?"
- "Which customers are at risk of churning?"
- "Which traffic source converts best to signups?"
- "Plan next week's X posts based on which channels brought paying customers last month."
- "What did we refund in Q1, and which plan has the worst refund ratio?"
Tools
All 19 tools are read-only and scoped to the accounts and sites your login owns.
| Area | Tools |
|---|---|
| Everything at once | get_business_snapshot, get_daily_snapshot |
| Revenue | get_mrr, get_mrr_by_plan, get_mrr_by_country, get_revenue_breakdown, get_refund_summary, get_goal_progress |
| Customers and churn | get_churn_rate, get_new_subscribers, list_at_risk_customers, list_top_customers, get_customer_concentration |
| Site traffic | list_tracked_sites, get_site_traffic, get_traffic_sources, get_monthly_events_usage |
| Marketing → revenue | get_revenue_by_source |
| Accounts | list_accounts |
Privacy
Customers are identified by payment-processor ID only. Names and emails never leave Chartsy. Access is checked on every call, so revoking a token or ending a subscription takes effect immediately. Policy: https://chartsy.app/privacy-policy
Repo layout
| Path | Used by |
|---|---|
skills/chartsy/SKILL.md | All skill-aware agents (single source of truth) |
skills/chartsy-growth-loop/SKILL.md | Weekly revenue-driven marketing loop: which channels brought paying customers, then UTM-tagged post drafts |
.claude-plugin/, .mcp.json | Claude Code plugin and marketplace |
.cursor-plugin/ | Cursor plugin and marketplace |
gemini-extension.json | Gemini CLI extension |
mcp.json | Generic mcpServers config for other clients |
Keep version in step across .claude-plugin/, .cursor-plugin/ and
gemini-extension.json when you release.
Support: [email protected]
License
Copyright © 2026 Chartsy. All rights reserved. See LICENSE. You may install these files to connect your own agent to Chartsy; no other rights are granted.
Comments
New to MCP?
From what the protocol is to installing your first server, these answers cover most first questions.
An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.