Affinity
Search your CRM, prep for meetings and update deal records from your AI assistant.
GoCardless API guidance plus live merchant access: payments, customers, mandates, payouts.
Opens Agent Studio, where connecting is one click. The connector URL below works in any MCP client.
Paste it into any MCP client. Setup docs
The GoCardless MCP server connects your GoCardless merchant account to Claude, ChatGPT or any MCP-compatible agent. It does two jobs: it teaches a model the GoCardless API so it can write integration code, and it reads and acts on your live account — payments, customers, mandates, payouts, refunds. GoCardless documents money-moving write actions here, not read-only lookups alone.
Verified connector
Listed by Anthropic as a partner connector in its Connectors Directory.
Connection checked by Agentman on .
Anthropic states this reflects the level of review a connector received, not a security audit.
Tool names from Anthropic's directory listing. This server requires sign-in, so we could not read tool descriptions or parameter schemas.
scopes_supported, so permissions are not machine-discoverable before sign-in.Yes. GoCardless documents write capabilities including creating payments, retrying and cancelling payments, issuing refunds, cancelling mandates, creating subscriptions and creating payment links. Money-moving actions display a preview and wait for your explicit confirmation before running. You choose which permissions to grant during sign-in, and read-only is one of the options offered.
Both, and the account access is real. GoCardless describes two capabilities: integration guidance covering endpoints, webhooks and code samples, and live account management. Once signed in, an agent can look up payments, customers, mandates, subscriptions, payouts, refunds and events on the environment you authorised, and take documented write actions on them.
GoCardless does not publish a per-tool schema, so the exact surface is undocumented. Its documentation enumerates the resource families an agent reaches: payments, customers, mandates, subscriptions, payouts, refunds and events. In GoCardless vocabulary those are API resources on your merchant account, not documentation pages, so the tool reads live merchant data rather than reference material.
You choose during sign-in. GoCardless runs two environments, Live for real payments and Sandbox for testing, and the connection journey asks which one you want. There is no in-session switch: changing environment means disconnecting and reconnecting. GoCardless recommends testing write actions in a free Sandbox account before running them against Live.
No. GoCardless states that bank account details are never shared with the AI tool, and that customer email addresses, phone numbers, post codes and identity numbers are masked before the model sees them. Customer records, mandates and payment history remain readable, so the connector still reaches data about identifiable payers even with masking applied.
Yes. The endpoint answered our anonymous probe with HTTP 401 and an OAuth challenge on 2026-08-19, so nothing is reachable without signing in. A free Sandbox account is enough for the integration-guidance side and for testing write actions. GoCardless also notes that free AI provider tiers typically do not support connectors.
https://developer.gocardless.com/developer-tools/mcp.md, the path named in GoCardless's own machine index) · retrieved 2026-08-19developer.gocardless.com/robots.txt allows all paths, names one sitemap, and carries no content-signal directive (fetched 2026-08-19) · retrieved 2026-08-19initialize POST to https://mcp.gocardless.com returns HTTP 401 with a www-authenticate Bearer challenge naming a resource_metadata URL. No authentication was attempted and no tool was called. · retrieved 2026-08-19authorization_servers and no scopes_supported) · retrieved 2026-08-19S256, dynamic registration, no scopes_supported) · retrieved 2026-08-19Read only, no prompts listed) · retrieved 2026-08-16Connect GoCardless once and your agents call these tools on their own: on a schedule, in a workflow, with nobody at the keyboard.