Add Tools

Add Tools & Grant Permissions

After adding any MCP server to your agent, you MUST grant the app access in databricks.yml.

Without this, you'll get permission errors when the agent tries to use the resource.

Workflow

Step 1: Add MCP server in src/mcp-servers.ts:

import { DatabricksMCPServer } from "@databricks/langchainjs";

export function getMCPServers(): DatabricksMCPServer[] {
  return [
    // Formula 1 Genie Space
    DatabricksMCPServer.fromGenieSpace("01f1037ebc531bbdb27b875271b31bf4"),

    // Add more MCP servers here...
  ];
}

Step 2: Grant access in databricks.yml:

resources:
  apps:
    agent_langchain_ts:
      resources:
        - name: 'f1_genie_space'
          genie_space:
            name: 'Formula 1 Race Analytics'
            space_id: '01f1037ebc531bbdb27b875271b31bf4'
            permission: 'CAN_RUN'

Step 3: Deploy with databricks bundle deploy (see deploy skill)

Resource Type Examples

See the examples/ directory for complete YAML snippets:

Custom MCP Servers (Databricks Apps)

Apps are not yet supported as resource dependencies in databricks.yml. Manual permission grant required:

Step 1: Get your agent app's service principal:

databricks apps get <your-agent-app-name> --output json | jq -r '.service_principal_name'

Step 2: Grant permission on the MCP server app:

databricks apps update-permissions <mcp-server-app-name> \
  --service-principal <agent-app-service-principal> \
  --permission-level CAN_USE

See examples/custom-mcp-server.md for detailed steps.

TypeScript-Specific Patterns

Adding Multiple MCP Servers

Edit src/mcp-servers.ts:

export function getMCPServers(): DatabricksMCPServer[] {
  const servers: DatabricksMCPServer[] = [];

  // Genie Space
  servers.push(
    DatabricksMCPServer.fromGenieSpace("01f1037ebc531bbdb27b875271b31bf4")
  );

  // SQL MCP
  servers.push(
    new DatabricksMCPServer({
      name: "dbsql",
      path: "/api/2.0/mcp/sql",
    })
  );

  // UC Functions
  servers.push(
    DatabricksMCPServer.fromUCFunction("main", "default")
  );

  // Vector Search
  servers.push(
    DatabricksMCPServer.fromVectorSearch("main", "default", "my_index")
  );

  return servers;
}

LangChain Agent Pattern

The agent uses standard LangGraph createReactAgent API:

// In src/agent.ts - uses standard LangGraph pattern
import { createReactAgent } from "@langchain/langgraph/prebuilt";

export async function createAgent(config: AgentConfig = {}) {
  // Load tools (basic + MCP if configured)
  const tools = await getAllTools(config.mcpServers);

  // Create agent using standard LangGraph API
  // Automatically handles tool execution, reasoning, and state management
  const agent = createReactAgent({
    llm: model,
    tools,
  });

  return new StandardAgent(agent, systemPrompt);
}

MCP Tool Types

Troubleshooting

See Troubleshooting Guide for common issues.

Quick tips:

• Permission errors: Check databricks.yml and redeploy
• Tool not found: Verify src/mcp-servers.ts and restart server
• MCP issues: See mcp-known-issues.md and mcp-best-practices.md in this directory

Additional Resources

• mcp-known-issues.md - Known MCP integration issues and status
• mcp-best-practices.md - Correct implementation patterns for MCP tools
• examples/ - YAML configuration examples for all resource types

Important Notes

• MLflow experiment: Already configured in template, no action needed
• Multiple resources: Add multiple entries under resources: list
• Permission types vary: Each resource type has specific permission values
• Deploy after changes: Run databricks bundle deploy after modifying databricks.yml
• Genie spaces: Use CAN_RUN permission for Genie spaces
• Service principal: Deployed apps run as service principals and need explicit resource grants