Most agents guess at Dataverse: invented columns, malformed OData, blind updates. MCP Dataverse gives them the real data model and 100 tools to work with it, under the identity and permissions you already manage.
A Dataverse environment is a system of record. The server is designed around that fact.
Agents can read entity definitions, relationships, option sets and keys before they act, so column names are checked rather than guessed.
Sign in with Microsoft Entra ID as a user, an app registration or a managed identity. Dataverse security roles decide what is allowed; the server adds no privilege. Security and guardrails
Up to 1,000 operations in one batch request, with changesets for atomic groups and per-operation results.
VS Code Copilot, Claude Desktop, Cursor, Windsurf, Gemini CLI, Copilot Studio and Codex CLI. Anything that speaks MCP.
One command starts an interactive wizard. Node.js 20 or later is the only prerequisite.
Every tool returns a typed result with a summary and next-step suggestions, so the agent can chain operations without guessing.
{
"tool": "dataverse_query",
"parameters": {
"entityLogicalName": "account",
"select": [
"name",
"revenue",
"createdon"
],
"filter": "revenue gt 1000000",
"orderby": "revenue desc",
"top": 5
}
}{
"summary": "5 accounts found (revenue > $1M)",
"data": [
{
"name": "Contoso Ltd",
"revenue": 8500000,
"createdon": "2024-01-15"
},
{
"name": "Fabrikam Inc",
"revenue": 5200000,
"createdon": "2023-11-02"
},
// ... 3 more records
],
"suggestions": [
"Use dataverse_get to fetch full details",
"Try dataverse_execute_fetchxml for complex joins"
]
}Six workflows that used to mean portal tabs and one-off scripts.
Map tables, relationships, option sets and entity keys across any Dataverse org — without opening the portal or reading a single doc.
list_tablesget_table_metadataget_relationshipsget_option_setOData filter, FetchXML aggregate, full-text search, paginated retrieval — your AI writes the right query, runs it, and explains the results.
queryexecute_fetchxmlsearchretrieve_multiple_with_pagingUp to 1,000 create, update or delete operations in one atomic batch. Build data migration scripts at the speed of a conversation.
batch_executecreateupdateupsertassignList roles, inspect privilege sets, detect over-permissioned users, grant or revoke record access — your full RBAC model at a glance.
list_rolesget_role_privilegesassign_role_to_usercheck_record_accessPull trace logs and workflow execution history, then let AI diagnose the error, explain the call stack, and suggest the fix.
get_plugin_trace_logsget_workflow_trace_logslist_plugin_stepslist_workflowsCreate tables, add columns, configure lookups and relationships, set environment variables, publish customizations — all via chat.
create_tablecreate_attributecreate_relationshipset_environment_variablepublish_customizationsTested with
Pick an authentication mode, run the installer, and ask your first question about your own data.