Visier MCP Server
Provider-hosted MCP server that connects AI agents to Visier's governed people analytics, workforce metrics, and Vee AI assistant.
Visier MCP (officially the Visier Query MCP Server) is a provider-hosted Model Context Protocol endpoint that lets MCP-compatible clients such as Claude Desktop, Cursor, Gemini CLI, and custom LangChain or Agentforce agents query Visier's governed people data model. Rather than building custom REST integrations against Visier APIs, agents authenticate once via OAuth 2.0 and then call standardized tools to run metric queries, list queries, and metadata searches, or to ask Vee, Visier's conversational people analytics agent.
The server exposes two groups of tools. The Vee AI tools let agents pass natural language questions to Vee and retrieve sample prompts. The Data Out tools let agents run aggregate metric queries, run list (record-level) queries, and search the analytic model for objects, properties, metrics, dimensions, and dimension members. All requests inherit the calling user's Visier permissions, data masking, and audit policies.
The server is hosted by Visier at a per-tenant URL of the form https://{vanity_name}.app.visier.com/visier-query-mcp, uses HTTPS with Server-Sent Events as transport, and is available to Visier customers who have purchased Vee. As of release, the feature is in alpha and may change without deprecation notices.
Tools
| Tool | Description |
|---|---|
Ask Vee AI agent |
Send a natural language question to Visier's Vee conversational analytics agent and return its answer. |
Get sample Vee questions |
Return curated sample prompts an agent can use as starting points for Vee. |
Run an aggregate metric query |
Execute an aggregate metric query (e.g. headcount, turnover, hire rate) against Visier's analytic model. |
Run a list query |
Execute a record-level list query to return rows of analytic objects (e.g. employees, applicants). |
Search analytic objects |
Search the Visier analytic model for objects such as Employee, Applicant, or Requisition. |
Search properties for an analytic object |
List or search properties defined on a given analytic object. |
Search metrics |
Find metrics in the Visier model by name or keyword. |
Search dimensions for a metric |
List dimensions available for grouping or filtering a specific metric. |
Search dimensions for an analytic object |
List dimensions available on a specific analytic object. |
Search dimension members for filtering |
Retrieve members of a dimension (e.g. department names) to use as filter values. |
Prerequisites
- A Visier tenant with loaded data
- Vee (Visier's conversational AI) entitlement
- OAuth 2.0 confidential client credentials registered for your MCP client
- A user profile with the "Data (Read, API)", "Legacy API Restricted", and "Query MCP Server" capabilities (plus "Vee Core" for Vee tools)
Find your server URL
Your Visier MCP URL is per-tenant:
https://{vanity_name}.app.visier.com/visier-query-mcp
To find your vanity name, open Visier Studio's global workspace and go to Settings > Single Sign-On. The hostname in your service provider endpoint (e.g. https://jupiter.visier.com/VServer/auth) is your vanity name.
OAuth 2.0 endpoints
Visier MCP uses the OAuth 2.0 authorization code grant:
- Authorization URL:
https://{vanity_name}.app.visier.com/hr/oauth2/authorize - Token URL:
https://{vanity_name}.app.visier.com/hr/oauth2/token - The
resourceparameter on OAuth requests must include the MCP server URL
Access tokens are valid for 60 minutes; refresh tokens for 8 hours.
Client configuration
Example config (Gemini CLI style, adaptable to other MCP clients):
{
"mcpServers": {
"visierQueryMcpServer": {
"url": "https://{vanity_name}.app.visier.com/visier-query-mcp",
"type": "http",
"oauth": {
"enabled": true,
"clientId": "{client_id}",
"clientSecret": "{client_secret}",
"authorizationUrl": "https://{vanity_name}.app.visier.com/hr/oauth2/authorize",
"tokenUrl": "https://{vanity_name}.app.visier.com/hr/oauth2/token"
}
}
}
}
A reference LangChain client is published at github.com/visier/sample-langchain-mcp-client and shows headless authentication via VISIER_OAUTH_CLIENT_ID, VISIER_OAUTH_CLIENT_SECRET, and VISIER_MCP_SERVER_URL environment variables.
- HR leaders ask Claude for engagement, turnover, or headcount metrics broken down by department or location without building custom dashboards.
- Talent and workforce planning teams query Visier's analytic model to identify under or over-staffed groups and pull supporting lists of employees.
- Custom enterprise agents (LangChain, Agentforce, internal copilots) embed governed people data alongside CRM or finance data for cross-functional reporting.
- Analysts use Vee through an MCP client to draft people analytics narratives and follow up with structured aggregate and list queries.
- IT and people operations expose a single MCP endpoint to replace bespoke API integrations across multiple AI tools while keeping Visier's permissions and audit trail intact.
- "Ask Vee what our voluntary turnover rate was last quarter, broken down by department."
- "Run an aggregate metric query for headcount by location for the last 12 months."
- "Search Visier for metrics related to engagement and list which dimensions are available."
- "List all employees in the Engineering organization who joined in the last 6 months."
- "Show me sample Vee questions I can use to explore workforce diversity."
- Officially built and hosted by Visier, with OAuth 2.0 and full inheritance of existing user permissions, data masking, and audit policies.
- Covers both natural language (Vee) and structured query tools (aggregate, list, model search) in one endpoint.
- Works with any MCP-compatible client, including Claude Desktop, Cursor, Gemini CLI, and custom LangChain/Agentforce agents; a reference LangChain sample client is published.
- No custom REST integration needed per AI tool; the analytic model is discoverable through the search tools.
- In alpha: features may change in breaking ways without deprecation notices.
- Requires a Visier tenant plus a Vee entitlement; not available to non-customers.
- Subject to standard Visier API rate limits and reflects data only as of the latest data load, not real-time operational HR data.
- Workday or SAP SuccessFactors community MCP servers for HRIS-level employee data (less analytics-focused).
- Direct use of Visier's REST APIs via a generic HTTP MCP server when Vee is not licensed.
- Snowflake or BigQuery MCP servers querying a people data warehouse, if Visier's governed model is not required.