A Fleet agent can work through the queue, but it needs the account history behind each item. Add Pathbound as a remote MCP server and the agent can read the resolved customer profile before it drafts, classifies, or requests an update.
Fleet discovers Pathbound tools from the remote server and makes them available beside its built-in tools. Add Pathbound to one agent while testing, then move it to workspace integrations if several agents need the same customer context.
A UI setup for teams that operate agents in LangSmith rather than maintaining LangChain application code.
Per-user OAuth when each person should use their own Pathbound permissions, or a scoped bearer credential for a shared agent identity.
One-agent and workspace-wide configuration paths, controlled by LangSmith’s MCP server permissions.
The same resolved profiles and event timelines used by the framework integration, without copying customer exports into Fleet instructions.
Start by attaching Pathbound to one Fleet agent. Use automatic OAuth for user-scoped access. Use a bearer header only when the agent is meant to act through a shared Pathbound credential.
Open the agent in the Fleet inbox and select Edit Agent. You need the MCP Server Create permission; a workspace admin can grant it.
In Toolbox, select MCP. Name the server Pathbound and enter https://mcp.pathbound.ai/mcp.
Choose OAuth 2.1 (Auto), save the agent, and complete the Pathbound browser sign-in when prompted.
Ask the agent to read a known contact before enabling broader jobs. Review the discovered write tools separately from reads.
In LangSmith, open Fleet → Integrations and select + Custom MCP. You can also use Settings → MCP servers → Add server.
Name the server Pathbound and enter https://mcp.pathbound.ai/mcp.
Choose Headers. Add the key Authorization with the value Bearer YOUR_PATHBOUND_API_KEY. Use a scoped key created for this agent, not a personal key copied from another environment.
Save the server, attach the discovered tools to the intended agents, and test one read-only job before expanding access.
Once connected, these are real jobs LangSmith Fleet runs against your customer data. Each recipe is one tested prompt—paste it and go.
Research every trial signup from the last 24 hours. For each one, pull their profile and company details, read what they did on our site and in the product, and work out who they are and whether the account is a fit. Write a short research brief as a note on each contact (who they are, what they did, verdict, suggested opener), then give me the list ranked by how promising they look, with one line each on why.See the full recipe—Trial-signup research
More jobs in use cases, each with a tested prompt.
Start with one agent while testing its instructions and tool access. Add Pathbound through Fleet Integrations or workspace MCP settings when several reviewed agents need the same server configuration.
Use OAuth 2.1 (Auto) when each person should connect their own Pathbound account and inherit their own permissions. Fleet prompts the user to sign in through Pathbound’s OAuth flow.
Use a bearer header for a shared or unattended agent identity. Create a Pathbound REST API key with only the scopes that agent needs and store it in the Fleet MCP configuration. Anyone who can configure or invoke that integration may be able to exercise those scopes.
Adding a remote MCP server requires the MCP Server Create permission in LangSmith. Ask a workspace admin to grant the permission or add the server on the team’s behalf.
Only through write tools allowed by the Pathbound credential and connected source. Keep the first agent read-only, then enable individual write tools with explicit review or approval steps. Versioning and rollback are not live yet.
No. Fleet connects to remote MCP servers from the LangSmith UI. Use the LangChain and LangGraph setup when you own the agent application and want to load Pathbound tools in Python or TypeScript.
Sign up, connect one CRM or drop the events snippet, and point your MCP client at mcp.pathbound.ai/mcp.
