CData Veeva Model Context Protocol (MCP) Server Implementation

This repository details the construction of a read-only MCP server component designed specifically for interacting with Veeva systems.

:warning: Important Notice: This utility is strictly limited to data retrieval operations (read-only). For a comprehensive solution supporting full data manipulation capabilities (read, write, update, delete, and executing actions), as well as simplified initial configuration, please explore our beta offering: CData MCP Server for Veeva (beta).

Core Functionality

This read-only MCP Server was engineered to empower Large Language Models (LLMs), such as Claude Desktop, to execute queries against real-time data residing within Veeva. Its operation is underpinned by the CData JDBC Driver for Veeva.

The CData JDBC Driver functions by abstracting Veeva data structures into a conventional relational SQL paradigm. Our server then envelops this driver, exposing the Veeva information via a simplified MCP structure. Consequently, LLMs can interrogate the live dataset using conversational language, eliminating the prerequisite for writing explicit SQL commands.

Deployment Procedure

1. Acquire Source Code: Initiate by cloning this repository: bash git clone https://github.com/cdatasoftware/veeva-mcp-server-by-cdata.git cd veeva-mcp-server-by-cdata

2. Compile Artifact: Build the necessary executable JAR file: bash mvn clean install

   The resulting file will be named: CDataMCP-jar-with-dependencies.jar 3. Driver Installation: Procure and install the CData JDBC Driver tailored for your data source: https://www.cdata.com/drivers/veeva/download/jdbc 4. Driver Licensing: Activate the JDBC Driver: * Navigate to the driver's installation directory, typically located here: * (Windows) C:\Program Files\CData\CData JDBC Driver for Veeva\ * (Mac/Linux) /Applications/CData JDBC Driver for Veeva/ * Execute the licensing utility: java -jar cdata.jdbc.veevavault.jar --license * Input your identification details (name, email) and the license key (use "TRIAL" if applicable). 5. Connection String Definition: Configure the pathway to your data repository (e.g., Salesforce): * Execute the driver executable to launch the connection utility: java -jar cdata.jdbc.veevavault.jar

   • Define the connection parameters within the utility, execute "Test Connection", and validate success.

     OAuth Note: If the target system mandates OAuth, authentication will proceed via a web browser prompt.

   • Secure and retain the generated connection string for the next phase.
   • Configuration File Creation (.prp): Construct a property file (e.g., veeva.prp) defining the server settings based on the following mandatory parameters:
   • Prefix: Identifier prefix for exposed tools.
   • ServerName: Assigned name for this specific MCP server instance.
   • ServerVersion: Version designation for the server.
   • DriverPath: Absolute path pointing to the installed JDBC driver JAR file.
   • DriverClass: The fully qualified name of the JDBC Driver Class (e.g., cdata.jdbc.veevavault.VeevaVaultDriver).
   • JdbcUrl: The connection string obtained in the previous step.
   • Tables: Specify individual table names separated by commas, or leave empty to enable access to all available data structures. env Prefix=veevault ServerName=VeevaDataGateway ServerVersion=1.0.1 DriverPath=ABSOLUTE/PATH/TO/cdata.jdbc.veevavault.jar DriverClass=cdata.jdbc.veevavault.VeevaVaultDriver JdbcUrl=jdbc:veevavault:InitiateOAuth=GETANDREFRESH;...other_params... Tables=

Integrating with AI Clients (e.g., Claude Desktop)

1. Client Configuration Update: Create or modify the client's configuration file (e.g., claude_desktop_config.json) to register this new MCP endpoint. Add an entry within the mcpServers object, utilizing the format below:

   Windows Environment

   { "mcpServers": { "veevault_gateway": { "command": "C:\PATH\TO\java.exe", "args": [ "-jar", "C:\PATH\TO\CDataMCP-jar-with-dependencies.jar", "C:\PATH\TO\veeva.prp" ] }, ... } }

   Linux/macOS Environment

   { "mcpServers": { "veevault_gateway": { "command": "/usr/bin/java", "args": [ "-jar", "/opt/mcp/CDataMCP-jar-with-dependencies.jar", "/etc/configs/veeva.prp" ] }, ... } }

   If the configuration file resides in a default location, copy it over: Windows bash cp C:\LOCAL\PATH\claude_desktop_config.json %APPDATA%\Claude\claude_desktop_config.json

   Linux/Mac bash cp /LOCAL/PATH/claude_desktop_config.json /Users/{user}/Library/Application\ Support/Claude/claude_desktop_config.json

2. Client Restart: Ensure the client application (e.g., Claude Desktop) is completely restarted or refreshed for the newly defined MCP services to appear in the available toolset.

Direct Server Execution

1. To initiate the MCP server independently (useful for direct JSON-RPC testing): bash java -jar /ABSOLUTE/PATH/TO/CDataMCP-jar-with-dependencies.jar /ABSOLUTE/PATH/TO/veeva.prp

   Constraint: The server communicates exclusively via standard input/output (stdio), mandating that any consuming client must execute on the same host machine as the server process.

Operational Usage Model

Once the MCP server is successfully linked via the client's configuration, the AI assistant gains access to integrated functionality for manipulating the underlying database objects. Typically, explicit tool invocation is unnecessary; simply pose data-centric inquiries in natural language. Examples of effective prompts include: * "Analyze the correlation coefficient between opportunities marked as 'Closed Won' and the assigned Account Industry classification." * "Quantify the total number of active support incidents recorded within the 'SUPPORT' operational project." * "List my scheduled calendar entries for the current day."

Exposed Tool Interface Reference

Tool names will incorporate the prefix defined in your configuration (e.g., veevault becomes {servername} in the examples below). * {servername}_get_tables: Fetches a comprehensive manifest of accessible data tables. For field introspection, chain this with the {servername}_get_columns tool. Output is delivered in CSV format, with the initial row detailing column headers. * {servername}_get_columns: Returns the schema definition (column list) for a specified table. Always reference {servername}_get_tables first to identify valid table names. Output adheres to CSV formatting, starting with header definitions. * {servername}_run_query: Permits the direct execution of arbitrary SQL SELECT statements against the data source.

JSON-RPC Invocation Schemas

For programmatic interaction bypassing an integrated AI client, utilize JSON-RPC 2.0 compliant payloads to invoke the available methods:

Invoking veevault_get_tables

{ "jsonrpc": "2.0", "id": 101, "method": "tools/call", "params": { "name": "veevault_get_tables", "arguments": {} } }

Invoking veevault_get_columns

{ "jsonrpc": "2.0", "id": 102, "method": "tools/call", "params": { "name": "veevault_get_columns", "arguments": { "table": "Account_Object__c" } } }

Invoking veevault_run_query

{ "jsonrpc": "2.0", "id": 103, "method": "tools/call", "params": { "name": "veevault_run_query", "arguments": { "sql": "SELECT Name, CreatedDate FROM [Account_Object__c] WHERE IsDeleted = true LIMIT 10" } } }

Common Issues and Resolutions

1. Server Visibility Failure: If the CData MCP Server does not appear in the client interface, confirm that the client application has been completely terminated (use Task Manager/Activity Monitor to verify no background processes remain).
2. Data Retrieval Errors: If the client cannot fetch data, meticulously re-verify the connection settings. Use the integrated Connection String builder to confirm validity, then ensure the resulting string is accurately transferred to the configuration (.prp) file.
3. Source Connection Problems: For connectivity failures to the primary data repository, direct consultation with the CData Support Team is recommended.
4. MCP Server Operation Feedback: Should you encounter issues specifically with the MCP server's functionality or wish to provide general commentary, engage with the CData Community.

Licensing

This MCP server component is distributed under the permissive MIT License. This grants users the freedom to employ, modify, and disseminate the software, provided compliance is maintained with the stipulations outlined in the MIT License text, accessible via the LICENSE file within this repository.

Supported Data Connectors Overview

A vast array of data systems are compatible via the CData JDBC infrastructure. While the focus here is Veeva, this table illustrates the breadth of supported sources:

Access	Act CRM	Act-On	Active Directory
ActiveCampaign	Acumatica	Adobe Analytics	Adobe Commerce
ADP	Airtable	AlloyDB	Amazon Athena
Amazon DynamoDB	Amazon Marketplace	Amazon S3	Asana
Authorize.Net	Avalara AvaTax	Avro	Azure Active Directory
Azure Analysis Services	Azure Data Catalog	Azure Data Lake Storage	Azure DevOps
Azure Synapse	Azure Table	Basecamp	BigCommerce
BigQuery	Bing Ads	Bing Search	Bitbucket
Blackbaud FE NXT	Box	Bullhorn CRM	Cassandra
Certinia	Cloudant	CockroachDB	Confluence
Cosmos DB	Couchbase	CouchDB	CSV
Cvent	Databricks	DB2	DocuSign
Dropbox	Dynamics 365	Dynamics 365 Business Central	Dynamics CRM
Dynamics GP	Dynamics NAV	eBay	eBay Analytics
Elasticsearch	Email	EnterpriseDB	Epicor Kinetic
Exact Online	Excel	Excel Online	Facebook
Facebook Ads	FHIR	Freshdesk	FTP
GitHub	Gmail	Google Ad Manager	Google Ads
Google Analytics	Google Calendar	Google Campaign Manager 360	Google Cloud Storage
Google Contacts	Google Data Catalog	Google Directory	Google Drive
Google Search	Google Sheets	Google Spanner	GraphQL
Greenhouse	Greenplum	HarperDB	HBase
HCL Domino	HDFS	Highrise	Hive
HubDB	HubSpot	IBM Cloud Data Engine	IBM Cloud Object Storage
IBM Informix	Impala	Instagram	JDBC-ODBC Bridge
Jira	Jira Assets	Jira Service Management	JSON
Kafka	Kintone	LDAP	LinkedIn
LinkedIn Ads	MailChimp	MariaDB	Marketo
MarkLogic	Microsoft Dataverse	Microsoft Entra ID	Microsoft Exchange
Microsoft OneDrive	Microsoft Planner	Microsoft Project	Microsoft Teams
Monday.com	MongoDB	MYOB AccountRight	MySQL
nCino	Neo4J	NetSuite	OData
Odoo	Office 365	Okta	OneNote
Oracle	Oracle Eloqua	Oracle Financials Cloud	Oracle HCM Cloud
Oracle Sales	Oracle SCM	Oracle Service Cloud	Outreach.io
Parquet	Paylocity	PayPal	Phoenix
PingOne	Pinterest	Pipedrive	PostgreSQL
Power BI XMLA	Presto	Quickbase	QuickBooks
QuickBooks Online	QuickBooks Time	Raisers Edge NXT	Reckon
Reckon Accounts Hosted	Redis	Redshift	REST
RSS	Sage 200	Sage 300	Sage 50 UK
Sage Cloud Accounting	Sage Intacct	Salesforce	Salesforce Data Cloud
Salesforce Financial Service Cloud	Salesforce Marketing	Salesforce Marketing Cloud Account Engagement	Salesforce Pardot
Salesloft	SAP	SAP Ariba Procurement	SAP Ariba Source
SAP Business One	SAP BusinessObjects BI	SAP ByDesign	SAP Concur
SAP Fieldglass	SAP HANA	SAP HANA XS Advanced	SAP Hybris C4C
SAP Netweaver Gateway	SAP SuccessFactors	SAS Data Sets	SAS xpt
SendGrid	ServiceNow	SFTP	SharePoint
SharePoint Excel Services	ShipStation	Shopify	SingleStore
Slack	Smartsheet	Snapchat Ads	Snowflake
Spark	Splunk	SQL Analysis Services	SQL Server
Square	Stripe	Sugar CRM	SuiteCRM
SurveyMonkey	Sybase	Sybase IQ	Tableau CRM Analytics
Tally	TaxJar	Teradata	Tier1
TigerGraph	Trello	Trino	Twilio
Twitter	Twitter Ads	Veeva CRM	Veeva Vault
Wave Financial	WooCommerce	WordPress	Workday
xBase	Xero	XML	YouTube Analytics
Zendesk	Zoho Books	Zoho Creator	Zoho CRM
Zoho Inventory	Zoho Projects	Zuora	... Many More