generated: '2026-08-02'
method: searched
status: published
source: https://docs.visier.com/developer/agents/mcp/mcp-server.htm
docs:
- https://docs.visier.com/developer/agents/mcp/mcp-server.htm
- https://docs.visier.com/developer/agents/mcp/mcp-server-set-up.htm
server:
  name: Visier Query MCP Server
  transport: http
  url: https://{vanity_name}.app.visier.com/visier-query-mcp
  streaming: Server-Sent Events (SSE) over HTTPS
  tenant_scoped: true
  note: >-
    The server URL is tenant-scoped. {vanity_name} is the customer's Visier vanity name, found under
    Settings > Single Sign-On (for example, a service provider endpoint of
    https://jupiter.visier.com/VServer/auth means the vanity name is "jupiter"). There is no
    anonymous/global MCP endpoint, so a live tools/list introspection could not be performed.
authentication:
  type: oauth2
  grants:
  - authorization_code
  - urn:ietf:params:oauth:grant-type:saml2-bearer
  access_token_lifetime: 60 minutes
  refresh_token_lifetime: configurable
  resource_parameter: >-
    OAuth requests must include the `resource` parameter set to the Visier Query MCP server URL.
  required_capabilities:
  - Query MCP Server
  - Vee Core (required for the Vee tools)
  - OAuth 2.0 Client Registration (administrators, to register the client)
introspection:
  tools_list_probed: false
  reason: >-
    The endpoint is tenant-scoped and OAuth-gated; no anonymous host resolves. Tool input schemas
    require authenticated tools/list introspection against a customer tenant. The tool list below is
    taken verbatim from Visier's published documentation (display names, not wire-level tool ids).
clients_documented:
- Claude Desktop
- Cursor
tools:
- name: Ask Vee AI agent
  category: vee
  description: >-
    Ask a natural-language workforce question. Vee interprets the intent, determines how to query the
    tenant's Visier data, and returns a natural-language response.
- name: Get sample Vee questions
  category: vee
  description: Retrieve sample questions Vee is able to answer.
- name: Get an object reference graph
  category: data-query
  description: Retrieve the reference graph of related analytic objects.
- name: Run an aggregate metric query
  category: data-query
  description: Run an aggregate query over a Visier metric.
- name: Run a list query
  category: data-query
  description: Run a list (row-level detail) query.
- name: Search analytic objects
  category: data-query
  description: Search the analytic objects available in the tenant's data model.
- name: Search properties for an analytic object
  category: data-query
  description: Search the properties defined on a given analytic object.
- name: Search metrics
  category: data-query
  description: Search the metrics available in the tenant's data model.
- name: Search dimensions for a metric
  category: data-query
  description: Search the dimensions available for a given metric.
- name: Search dimensions for an analytic object
  category: data-query
  description: Search the dimensions available for a given analytic object.
- name: Search dimension members for filtering
  category: data-query
  description: Search dimension members, for building filters on a query.
prompts:
- name: Employee analysis instructions
  description: >-
    Instructions to analyze data on a specific employee — jobs, organizations, tenure, compensation,
    performance, risk of resignation, and related analytic objects, optionally compared against
    organization-wide metrics.
- name: Candidate analysis instructions
  description: >-
    Instructions to analyze data on a specific candidate — job, employer, location, skills, and
    related analytic objects.
- name: Requisition analysis instructions
  description: >-
    Instructions to analyze and compare the applicants for a specific requisition — job, location,
    reason, status, and related analytic objects.
limitations:
- Results reflect the tenant's data as of the latest data load in Visier.
- Visier API rate limits apply to MCP calls (100 calls/minute, 6000 calls/hour); calls over the limit are rejected.
- Natural-language understanding depends on the connected AI client's model.
security_considerations:
- >-
  Visier documents lethal-trifecta / indirect prompt-injection risk and advises against connecting
  untrusted MCP servers to agents that also access Visier MCP.
- >-
  The customer is the controller of the decision to send Visier data to third-party AI tools; use of
  those tools is governed by the customer's own agreement with the tool provider.
x-evidence:
  fetched: '2026-08-02'
  urls:
  - https://docs.visier.com/developer/agents/mcp/mcp-server.htm
  - https://docs.visier.com/developer/agents/mcp/mcp-server-set-up.htm
  http_status: 200