n8n runs your automations on a schedule or a trigger. Add Pathbound as a remote MCP server and an n8n AI Agent can read the contact, company, timeline, and recent activity behind a new signup, a quiet account, or a support event before it acts.
n8n is built to move data between the tools you run. Pathbound gives an n8n AI Agent one resolved customer record to read first, with identities matched across the sources you have connected, through one remote MCP server. MCP is the open protocol agents use to call tools.
A new signup, a stage change, or a nightly run starts a workflow that already knows the account behind the row that fired it.
Pathbound matches CRM records, support history, email, meetings, and website events to the same contact and company before the agent requests them.
A scoped API key sets what the agent can read. Start with read-only contact, company, and event access; add write scopes only when a workflow needs them.
This path uses n8n’s built-in MCP Client Tool, the sub-node you attach to an AI Agent node. No community node is required.
In Pathbound, open REST API keys and create a key with contacts:read, companies:read, and events:read. Leave write and action scopes off.
In n8n, create a Bearer Auth credential and paste the key as the token. n8n keeps it in its credential store, apart from the workflow itself.
sk_your_api_keyOn your AI Agent node, add an MCP Client Tool. Set Server Transport to HTTP Streamable, enter the Pathbound endpoint, and select the Bearer credential. Set Tools to Include to the read tools the workflow needs.
https://mcp.pathbound.ai/mcpExecute the node with one account you can check. A prompt like “Summarize Acme from Pathbound and list everything from the last seven days, with the source and date” should return the record. Confirm it matches your dashboard before you put the workflow on a schedule.
n8n’s MCP Client Tool reference covers transport, authentication, and tool selection. Use a recent n8n version — HTTP Streamable is the current transport and SSE is deprecated.
Each one runs an n8n AI Agent over the Pathbound record, so the automation reads the resolved customer before it decides what to do.
A webhook or form trigger starts an agent that reads the new contact’s resolved profile and recent activity, then adds an enriched note when the key allows a write.
See the trial-signup recipe →A schedule trigger runs an agent over identified accounts with dropping activity or an open ticket, then posts a short risk list to Slack or email.
See the quiet-account recipe →When a ticket event fires, an agent pulls the contact’s conversations and last promises so the reply does not ask the customer to repeat themselves.
See the full-context support recipe →More patterns live in the use-case recipes. Start read-only, then add a single write step once the read side is right.
An n8n workflow can run without a person watching. Treat what the agent can read and what it can change as separate decisions.
Use the smallest API key scope that serves the job. Rotate or revoke the key from the Pathbound dashboard. A key without actions:write receives no action tools.
Store the key as a credential, never in workflow parameters. Limit who can edit the workflow and its credentials, and keep the first version read-only before you schedule it.
Yes. n8n’s MCP Client Tool consumes a remote MCP server. Point it at https://mcp.pathbound.ai/mcp over HTTP Streamable, authenticate with a scoped Pathbound API key, and attach it to your AI Agent node.
No. The MCP Client Tool ships with n8n, and Pathbound already exposes its tools over MCP. Older community nodes exist, but the built-in tool covers this connection.
Use HTTP Streamable. Pathbound serves the current Streamable HTTP transport, and n8n’s SSE option is deprecated. If the agent reports connection errors or a retry loop while HTTP Streamable is selected, update n8n — some 2026 builds shipped a transport-selection bug — and confirm the endpoint ends in /mcp.
A scoped key is predictable for an unattended workflow and can stay read-only. The MCP Client Tool also supports OAuth2 when each run should carry a person’s own Pathbound identity and permissions.
Only when the key includes the required write scopes and the connected source allows the tool. Keep the first workflow read-only. Governed write-back is rolling out; versioning and rollback are not live yet.
Sign up, connect one CRM or drop the events snippet, and point your MCP client at mcp.pathbound.ai/mcp.