Chartsy is a hosted, read-only MCP server that gives Make scenarios your MRR, churn and refunds from Stripe, Paddle, Polar, Lemon Squeezy, Creem or Freemius, plus site traffic and revenue by marketing channel. You connect it through Make's MCP Client app with a personal token. Last updated October 2026.
Using Chartsy, summarise yesterday in 3 short bullets for Slack. Say whether each number was normal against the 7-day average.
Sent to Slack by the next module:
Scenarios run without you, so Make connects with a personal token instead of a sign-in.
https://mcp.chartsy.app/mcp
Test it first with Call a tool and the list_tracked_sites tool: it takes no arguments, so a result means the connection works.
Using Chartsy, summarise yesterday in 3 short bullets for Slack. Say whether each number was normal against the 7-day average.
Also automating with n8n or Zapier? See Chartsy + n8n and Chartsy + Zapier.
All three use the same connection. They differ in who decides which Chartsy tool runs.
| Surface | How it works | ChartsyAI questions | Best for |
|---|---|---|---|
Call a toolThis page | You pick the tool and map its arguments | Exactly 1 per run, no AI model | Sheets, alerts and fixed weekly reports |
Execute an action with AI | You describe a task; a model picks the tools | Usually 1 or 2 per run, set by the model | Written summaries and open questions |
MCP Tools | Hands Chartsy tools to a Make AI Agent | One per tool the agent calls | Agents that answer team questions on demand |
Tasks for Execute an action with AI, and fixed tool calls. A model picks its own tools, so the count can vary.
Using Chartsy, summarise yesterday in 3 short bullets for Slack. Say whether each number was normal against the 7-day average.
Posted to #numbers:
Weekdays only: about 22 ChartsyAI questions a month.
Call a tool: get_revenue_by_source for last week, by source.
Rows added to Google Sheets:
One call a week, no AI model: about 4 ChartsyAI questions a month.
Using Chartsy, write Monday's team email: MRR, new subscribers and churn for last week, and the one thing that changed.
Sent with Gmail:
Sample figures for a made-up business.
Call a tool: list_at_risk_customers.
3 rows added to the follow-up sheet:
IDs only: names and emails stay in Chartsy.
Using Chartsy, report progress on each goal and flag any that are behind pace.
Posted every Friday:
Sample figures for a made-up business.
More in the MCP prompt library.
Call a tool runs get_revenue_by_source directly and an Iterator turns each channel into a sheet row. Over a few weeks the sheet shows which channels keep bringing customers who pay.
by to campaign or content to compare UTM campaigns or posts1. Schedule
Every Monday at 8:00
2. MCP Client: Call a tool
Tool Name: get_revenue_by_source
start_date: {{formatDate(addDays(now; -7); "YYYY-MM-DD")}}
end_date: {{formatDate(addDays(now; -1); "YYYY-MM-DD")}}
by: source
3. JSON: Parse JSON
(if the tool result arrives as text)
4. Iterator
Array: groups
5. Google Sheets: Add a row
Week · Source · Paying customers · MRRSignups only tell you half the story. Make can now follow each channel into your billing data, so you see which ones bring customers who pay and stay.
Needs the Chartsy tracking script on your site and a connected payment provider. Groups under 5 paying customers are flagged as inconclusive, and retention needs about 90 days to mean much.
Which channels brought paying customers last quarter, and what are they worth now?
Search brings the most money; the newsletter converts best.
The same numbers as your Chartsy dashboard. In Execute an action with AI, the Tools field limits what the model can call.
get_mrrget_mrr_by_planget_mrr_by_countryget_revenue_breakdownget_refund_summaryget_goal_progressget_churn_rateget_new_subscriberslist_at_risk_customerslist_top_customersget_customer_concentrationlist_tracked_sitesget_site_trafficget_traffic_sourcesget_monthly_events_usageget_revenue_by_sourceget_business_snapshotget_daily_snapshotlist_accountsNumbers and customer IDs reach your scenario. Names, emails and write access never do.
Make can read your numbers. It can't change Chartsy, your payment provider's data, or your customers.
Customers are payment-processor IDs; names and emails stay in Chartsy. Make sees only the accounts and sites your login owns, checked on every call.
Each call uses one ChartsyAI question from your monthly allowance, failed calls included. Broad questions usually take one call.
Most problems are the token, the URL or the arguments.
Check the URL is exactly https://mcp.chartsy.app/mcp and the token is in API Key / Access token with no spaces. Make's MCP Client doesn't offer a transport setting; if the connection still fails, tell us at support@chartsy.app with the error Make shows.
The token is wrong, revoked or expired. Generate a new one in Chartsy → Settings → Claude & MCP and update the connection in Make.
Dates must be YYYY-MM-DD. Map them with formatDate, for example formatDate(addDays(now; -7); "YYYY-MM-DD") for a week ago.
Connect Stripe, Paddle, Polar, Lemon Squeezy, Creem or Freemius in Chartsy and let the first sync finish. The demo data added at signup doesn't count.
get_revenue_by_source needs the Chartsy tracking script on your site and signups from the chosen dates. Run list_tracked_sites to check the site is tracked.
Each tool call uses one ChartsyAI question, failed calls included. When the monthly allowance is used up, calls stop until it resets on the 1st, or until you upgrade your Chartsy plan. Scenarios that run every few minutes use it up fast.
Short answers for people building Make scenarios.
Generate a personal token in Chartsy → Settings → Claude & MCP. In Make, add the MCP Client app to a scenario, create a connection with + New MCP Server, enter https://mcp.chartsy.app/mcp as the URL and paste the token into API Key / Access token.
Only for Execute an action with AI, which runs on Make's AI provider or your OpenAI connection. Call a tool runs one Chartsy tool directly, with no model, and passes the result to the next module.
One per tool call, failed calls included. A weekday Slack summary with one call is about 22 questions a month; a weekly sheet is about 4. Execute an action with AI may call more than one tool per run, so schedule scenarios sensibly.
Connecting costs nothing extra on the Chartsy side. It works on any Chartsy plan, including the free trial, and each tool call uses one ChartsyAI question. Make's own operations and AI credits are billed by Make.
Make reads whatever you have connected to Chartsy: Stripe, Paddle, Polar, Lemon Squeezy, Creem, Freemius or a custom source through the Chartsy API.
No. All 19 Chartsy tools are read-only. A scenario can't change Chartsy, your payment provider's data, or your customers.
Delete the Chartsy connection under Connections in Make. To cut access everywhere, revoke the token in Chartsy under Settings → Claude & MCP; it stops working on the next call.
Yes. The get_revenue_by_source tool follows the people who signed up from each source, medium, UTM campaign or post into your billing data, and returns how many became paying customers, what they're worth in MRR today, how many churned and the retention rate. You can choose first touch (where they first found you) or last touch (what made them sign up). It needs the Chartsy tracking script on your site and a connected payment provider. Groups with fewer than 5 paying customers are flagged as too small to call.
Same server, same numbers. Or see all AI agents and the MCP server reference.
Connect your payment provider to Chartsy, add the MCP Client app, and schedule the scenario.
Product names are trademarks of their owners, used only to say what Chartsy works with. Chartsy isn't affiliated with or endorsed by them.
Ministry of Economy and Innovation
Startup AlbaniaThe Chartsy program is realized with the financial support of the Albanian Government through the Ministry of Economy and Innovation, under the Grant 2026 scheme, and is implemented by the Innovation4Albania Agency.