generated: '2026-09-16'
method: searched
source: https://docs.getmontecarlo.com/docs/mcp-server
status: gated
name: Monte Carlo MCP Server
description: >-
  First-party, fully hosted Model Context Protocol server for the Monte Carlo data + AI observability
  platform. Lets an agent investigate alerts, explore assets and lineage, create and tune monitors, and
  evaluate AI agent performance. Runs on AWS Lambda behind Monte Carlo's Integration Gateway, is stateless,
  and executes every call under the authenticated user's identity against Monte Carlo's GraphQL API.
  An Editor role or above is required.
server:
  name: monte-carlo
  transport: streamable-http
  url: https://mcp.getmontecarlo.com/mcp
deployment:
  mode: remote
  endpoint: https://mcp.getmontecarlo.com/mcp
  auth: oauth
  verified: probed
  probe: gated
  checked: '2026-09-16'
  probe_evidence:
    url: https://mcp.getmontecarlo.com/mcp
    method: POST tools/list (anonymous)
    http_status: 401
    www_authenticate: Bearer error="invalid_token", error_description="Authentication required",
      resource_metadata="https://mcp.getmontecarlo.com/.well-known/oauth-protected-resource/mcp",
      scope="offline_access"
  note: >-
    Remote endpoint. The docs' key-based setup runs `npx -y mcp-remote https://mcp.getmontecarlo.com/mcp`
    with an Authorization header; mcp-remote is a generic third-party stdio bridge to the same hosted
    endpoint, not a Monte Carlo package, so this is not a local-stdio deployment.
authentication:
  - type: oauth2.1
    detail: OAuth 2.1 with Dynamic Client Registration (no client ID/secret issued); PKCE S256.
    authorization_server: https://auth.getmontecarlo.com
    protected_resource_metadata: https://mcp.getmontecarlo.com/.well-known/oauth-protected-resource/mcp
    scopes: [offline_access]
    docs: https://docs.getmontecarlo.com/docs/mcp-server#option-2-oauth-recommended-for-all-other-agents
  - type: api-key
    detail: >-
      MCP Server keys (key type "MCP server") only; standard API keys are rejected. Sent as
      `Authorization: Bearer <KEY_ID>:<KEY_SECRET>` or as `x-mcd-id` / `x-mcd-token` headers.
    create: montecarlo mcp create-key --description "..." (CLI v0.141.4+), or GraphQL mutation createMcpIntegrationKey
    docs: https://docs.getmontecarlo.com/docs/mcp-server#option-3-api-keys-advanced
endpoints:
  - url: https://mcp.getmontecarlo.com/mcp
    purpose: Primary MCP endpoint documented for all clients (Claude connector, OAuth clients, mcp-remote).
    source: https://docs.getmontecarlo.com/docs/mcp-server
    anonymous_probe: 401 (OAuth challenge)
  - url: https://mcp.getmontecarlo.com/mcp/toolkit
    purpose: Endpoint the first-party Agent Toolkit plugins configure (Claude Code, Codex, Cortex Code, Cursor).
    source: https://github.com/monte-carlo-data/mc-agent-toolkit/blob/main/plugins/claude-code/.mcp.json
    anonymous_probe: 401 (OAuth challenge; resource metadata at /.well-known/oauth-protected-resource/mcp/toolkit)
  - url: https://montecarlo.ai/wp-json/mcp/mcp-oauth-server
    purpose: >-
      Separate MCP endpoint advertised by the montecarlo.ai marketing site's
      /.well-known/oauth-protected-resource (WordPress MCP adapter, scope "mcp"). Not mentioned in the
      developer docs; its tool surface is unknown.
    source: https://montecarlo.ai/.well-known/oauth-protected-resource
    anonymous_probe: 401 {"code":"mcp_unauthorized"}
distribution:
  - channel: Claude Connectors Directory
    url: https://claude.com/connectors/monte-carlo
    note: Anthropic-verified connector, "Made by Monte Carlo", connector URL https://mcp.getmontecarlo.com/mcp, added May 2026.
  - channel: Snowflake Cortex Agents external MCP server
    url: https://docs.getmontecarlo.com/docs/mcp-server#connect-from-a-snowflake-cortex-agent
toolset_selection_header: x-mcd-toolset
toolsets:
  - name: default
    description: Always active. Alerts, monitors, assets, lineage, data operations, and utility tools.
  - name: extended
    description: Default tools plus use-case and dashboard-widget tools.
    adds: [get_use_cases, get_use_case_tables, get_use_case_table_summary]
  - name: agent_observability
    description: Default tools plus AI agent inspection tools.
    adds: [get_agent_metadata, get_agent_conversation, get_agent_trace]
request_controls_source: https://docs.getmontecarlo.com/docs/mcp-server-reference-and-operations
request_controls:
  - header: x-mcp-readonly
    effect: Exposes only read-only tools; write tools (update_alert, create_or_update_* monitors) are hidden and rejected.
  - header: x-mcp-tools
    effect: Comma-separated allow-list intersected with the active toolset; enforced at call time.
tools_source: >-
  Live tools/list is OAuth-gated, so no inputSchema was captured. Tool names below come from two published
  sources: the Claude Connectors Directory listing (25 tools, shown in camelCase display form) and the
  tool names Monte Carlo's own docs cite in snake_case. Schemas require authenticated introspection.
tools:
  - {name: get_alerts, directory_name: getAlerts, toolset: default, access: read, cited_in: [directory, docs]}
  - {name: update_alert, directory_name: updateAlert, toolset: default, access: write, cited_in: [directory, docs]}
  - {name: set_alert_owner, directory_name: setAlertOwner, toolset: default, access: write, cited_in: [directory]}
  - {name: create_or_update_alert_comment, directory_name: createOrUpdateAlertComment, toolset: default, access: write, cited_in: [directory]}
  - {name: search, directory_name: search, toolset: default, access: read, cited_in: [directory, docs]}
  - {name: get_table, directory_name: getTable, toolset: default, access: read, cited_in: [directory, docs]}
  - {name: get_queries_for_table, directory_name: getQueriesForTable, toolset: default, access: read, cited_in: [directory]}
  - {name: get_query_data, directory_name: getQueryData, toolset: default, access: read, cited_in: [directory]}
  - {name: get_asset_lineage, directory_name: getAssetLineage, toolset: default, access: read, cited_in: [directory, docs]}
  - {name: get_downstream_bi_reports, directory_name: getDownstreamBiReports, toolset: default, access: read, cited_in: [directory]}
  - {name: get_monitors, directory_name: getMonitors, toolset: default, access: read, cited_in: [directory, docs]}
  - {name: get_audiences, directory_name: getAudiences, toolset: default, access: read, cited_in: [directory]}
  - {name: create_validation_monitor_mac, directory_name: createValidationMonitorMac, toolset: default, access: write, cited_in: [directory]}
  - {name: create_metric_monitor_mac, directory_name: createMetricMonitorMac, toolset: default, access: write, cited_in: [directory]}
  - {name: create_comparison_monitor_mac, directory_name: createComparisonMonitorMac, toolset: default, access: write, cited_in: [directory]}
  - {name: create_custom_sql_monitor_mac, directory_name: createCustomSqlMonitorMac, toolset: default, access: write, cited_in: [directory]}
  - {name: get_use_cases, directory_name: getUseCases, toolset: extended, access: read, cited_in: [directory, docs]}
  - {name: get_use_case_table_summary, directory_name: getUseCaseTableSummary, toolset: extended, access: read, cited_in: [directory, docs]}
  - {name: get_use_case_tables, directory_name: getUseCaseTables, toolset: extended, access: read, cited_in: [directory, docs]}
  - {name: create_custom_dashboard_widget, directory_name: createCustomDashboardWidget, toolset: extended, access: write, cited_in: [directory]}
  - {name: test_connection, directory_name: testConnection, toolset: default, access: read, cited_in: [directory]}
  - {name: get_domains, directory_name: getDomains, toolset: default, access: read, cited_in: [directory]}
  - {name: get_user, directory_name: getUser, toolset: default, access: read, cited_in: [directory]}
  - {name: get_current_time, directory_name: getCurrentTime, toolset: default, access: read, cited_in: [directory]}
  - {name: get_validation_predicates, directory_name: getValidationPredicates, toolset: default, access: read, cited_in: [directory]}
  - {name: alert_assessment, toolset: default, access: read, cited_in: [docs]}
  - {name: create_or_update_table_monitor, toolset: default, access: write, cited_in: [docs]}
  - {name: get_agent_metadata, toolset: agent_observability, access: read, cited_in: [docs]}
  - {name: get_agent_conversation, toolset: agent_observability, access: read, cited_in: [docs]}
  - {name: get_agent_trace, toolset: agent_observability, access: read, cited_in: [docs]}
  - {name: create_agent_metric_monitor, toolset: agent_observability, access: write, cited_in: [docs]}
tools_note: >-
  snake_case names for directory-only tools are the directory's camelCase display names converted, not
  confirmed wire names; toolset and access for directory-only tools are inferred from the docs' toolset
  table and the tool verb. The Agent Toolkit skills reference further tool names (e.g. get_agent_traces,
  run_troubleshooting_agent, get_alert_agent_classification, get_etl_issues, analyze_storage_costs) that
  appear to be served on the /mcp/toolkit endpoint; they are not listed here because no tool list for that
  endpoint is published.
prompts:
  - {name: mc-asset-health, description: Check the current health of a table before querying it.}
  - {name: mc-triage, description: Triage your Monte Carlo alerts so you know where to focus.}
  - {name: mc-incident-response, description: 'Full incident orchestration, from triage through root cause to remediation.'}
  - {name: mc-proactive-monitoring, description: Discover coverage gaps and create monitors to close them.}
related:
  agent_toolkit: https://github.com/monte-carlo-data/mc-agent-toolkit
  technical_overview: https://docs.getmontecarlo.com/docs/monte-carlo-mcp-server-technical-overview
  reference: https://docs.getmontecarlo.com/docs/mcp-server-reference-and-operations
  security: https://docs.getmontecarlo.com/docs/ai-security
  crosswalk: mcp/monte-carlo-tool-crosswalk.yml