Give Cursor the customer record behind the work.

Investigate a reported issue beside the code, or run a customer workflow without a repository. Pathbound gives Cursor current profiles, timelines, conversations, and product activity through one MCP connection.

Where it runs

One customer connection across Cursor.

Cursor is useful beside a repository, but it is no longer limited to one. The same customer tools can support an interactive investigation or a recurring operational job.

Connect

Connect Pathbound to Cursor.

The one-click link installs the remote server definition. Use the manual path when you want to review or share the configuration yourself.

  1. Add the remote MCP server

    Select Add to Cursor above, or merge this entry into ~/.cursor/mcp.json. Use .cursor/mcp.json only when the server definition should travel with a specific repository.

    ~/.cursor/mcp.json
    {
      "mcpServers": {
        "pathbound": {
          "type": "http",
          "url": "https://mcp.pathbound.ai/mcp"
        }
      }
    }
  2. Complete Pathbound OAuth

    In Cursor, open the Pathbound MCP entry and authenticate. From Cursor Agent CLI, run the login command below. Your browser opens the Pathbound authorization flow; no API key needs to be pasted into an interactive configuration.

  3. Verify the connection

    Confirm that Pathbound is enabled and inspect the live tool catalog before asking it to retrieve a known customer.

    Terminal
    cursor-agent mcp login pathbound
    cursor-agent mcp list
    cursor-agent mcp list-tools pathbound

Cursor documents remote Streamable HTTP servers, OAuth, global and project configuration, and CLI MCP commands in its MCP reference and CLI reference.

Customer operations

Use Cursor as an investigation and operations workspace.

Pathbound resolves the customer identity before Cursor asks for it. Cursor can work from that record alongside the other tools your team has approved.

No-repo workflow

Start with a read-only customer review.

Create an Automation in Cursor, leave the repository field empty, enable Pathbound, and begin with a schedule you can review. Add external triggers only after the output is stable.

Cursor documents no-repo templates for customer health, product analytics, product FAQ, finance, and Slack digests in its Automations release notes.

Automation instruction
Every Monday, review customer activity from the last 7 days.

Use Pathbound to:
1. Compare activity by event type and company with the prior 7 days.
2. Retrieve the company and recent contact activity behind material changes.
3. Return the evidence, affected customers, and open questions.

Read only. Do not update customer records or send messages.
Access

Separate reading from acting.

Cursor asks for approval before running MCP tools by default in an interactive session. Keep that boundary in place while testing Pathbound, and keep a no-repo Automation read-only until its queries and output have been reviewed.

For a headless runner, create a Pathbound REST API key with only the required read scopes. Do not place the key in a versioned .cursor/mcp.json. Enable individual write or send tools only when the workflow has an explicit owner and approval path.

Configuration, teams, and scope.

Can Cursor use Pathbound without a code repository?

Yes. Cursor Automations can run with no repository attached. Add Pathbound as an MCP, choose it for the Automation, then give the workflow a schedule or trigger. This is useful for customer-health reviews, product-activity summaries, and support research.

Is the same Pathbound connection available in Cursor and Cursor Agent CLI?

Cursor Agent CLI reads the same mcp.json configuration as the desktop app. A global entry in ~/.cursor/mcp.json is available across workspaces on that machine. Team administrators can also distribute approved MCP servers through a team marketplace.

How does Cursor authenticate with Pathbound?

For an interactive session, add the Pathbound URL and complete OAuth from Cursor or with cursor-agent mcp login pathbound. For a headless workflow that cannot open a browser, use a scoped Pathbound REST API key as a bearer credential.

Can non-technical teams use this setup?

The recurring workflow can be written in plain language and run without a repository. An administrator should configure the MCP connection, credential, triggers, and allowed tools before sharing it with support, customer success, or operations teams.

Can Cursor change customer data through Pathbound?

Only when the Pathbound credential allows the relevant write tool. Start with read-only customer, company, conversation, and event access. Enable individual write paths only after the workflow and approval boundary are reviewed. Versioning and rollback are not live yet.

Related
Try it on your own data

Within five minutes, your agent is querying real customer context.

Sign up, connect one CRM or drop the events snippet, and point your MCP client at mcp.pathbound.ai/mcp.