skills/entra-agent-user/SKILL.md
Create Agent Users in Microsoft Entra ID from Agent Identities, enabling AI agents to act as digital workers with user identity capabilities in Microsoft 365 and Azure environments.
npx skillsauth add jyjeanne/ai-setup-forge entra-agent-userInstall this skill globally with one command. Works with Claude Code, Cursor, and Windsurf.
3 of 9 scanners reported clean
Some scanners were skipped, did not run, or reported a non-clean status. Review each row below.
An agent user is a specialized user identity in Microsoft Entra ID that enables AI agents to act as digital workers. It allows agents to access APIs and services that strictly require user identities (e.g., Exchange mailboxes, Teams, org charts), while maintaining appropriate security boundaries.
Agent users receive tokens with idtyp=user, unlike regular agent identities which receive idtyp=app.
ServiceIdentity) created from an agent identity blueprintAgentIdUser.ReadWrite.IdentityParentedBy (least privileged)AgentIdUser.ReadWrite.AllUser.ReadWrite.AllImportant: The
identityParentIdmust reference a true agent identity (created via an agent identity blueprint), NOT a regular application service principal. You can verify by checking that the service principal has@odata.type: #microsoft.graph.agentIdentityandservicePrincipalType: ServiceIdentity.
Agent Identity Blueprint (application template)
│
├── Agent Identity (service principal - ServiceIdentity)
│ │
│ └── Agent User (user - agentUser) ← 1:1 relationship
│
└── Agent Identity Blueprint Principal (service principal in tenant)
| Component | Type | Token Claim | Purpose |
|---|---|---|---|
| Agent Identity | Service Principal | idtyp=app | Backend/API operations |
| Agent User | User (agentUser) | idtyp=user | Act as a digital worker in M365 |
Before creating an agent user, confirm the agent identity is a proper agentIdentity type:
GET https://graph.microsoft.com/beta/servicePrincipals/{agent-identity-id}
Authorization: Bearer <token>
Verify the response contains:
{
"@odata.type": "#microsoft.graph.agentIdentity",
"servicePrincipalType": "ServiceIdentity",
"agentIdentityBlueprintId": "<blueprint-id>"
}
Connect-MgGraph -Scopes "Application.Read.All" -TenantId "<tenant>" -UseDeviceCode -NoWelcome
Invoke-MgGraphRequest -Method GET `
-Uri "https://graph.microsoft.com/beta/servicePrincipals/<agent-identity-id>" | ConvertTo-Json -Depth 3
Common mistake: Using an app registration's
appIdor a regular application service principal'sidwill fail. Only agent identities created from blueprints work.
POST https://graph.microsoft.com/beta/users/microsoft.graph.agentUser
Content-Type: application/json
Authorization: Bearer <token>
{
"accountEnabled": true,
"displayName": "My Agent User",
"mailNickname": "my-agent-user",
"userPrincipalName": "[email protected]",
"identityParentId": "<agent-identity-object-id>"
}
| Property | Type | Description |
|---|---|---|
| accountEnabled | Boolean | true to enable the account |
| displayName | String | Human-friendly name |
| mailNickname | String | Mail alias (no spaces/special chars) |
| userPrincipalName | String | UPN — must be unique in the tenant (alias@verified-domain) |
| identityParentId | String | Object ID of the parent agent identity |
Connect-MgGraph -Scopes "User.ReadWrite.All" -TenantId "<tenant>" -UseDeviceCode -NoWelcome
$body = @{
accountEnabled = $true
displayName = "My Agent User"
mailNickname = "my-agent-user"
userPrincipalName = "[email protected]"
identityParentId = "<agent-identity-object-id>"
} | ConvertTo-Json
Invoke-MgGraphRequest -Method POST `
-Uri "https://graph.microsoft.com/beta/users/microsoft.graph.agentUser" `
-Body $body -ContentType "application/json" | ConvertTo-Json -Depth 3
400 Bad Request.userPrincipalName must be unique. Don't reuse an existing user's UPN.Assigning a manager allows the agent user to appear in org charts (e.g., Teams).
PUT https://graph.microsoft.com/beta/users/{agent-user-id}/manager/$ref
Content-Type: application/json
Authorization: Bearer <token>
{
"@odata.id": "https://graph.microsoft.com/beta/users/{manager-user-id}"
}
$managerBody = '{"@odata.id":"https://graph.microsoft.com/beta/users/<manager-user-id>"}'
Invoke-MgGraphRequest -Method PUT `
-Uri "https://graph.microsoft.com/beta/users/<agent-user-id>/manager/`$ref" `
-Body $managerBody -ContentType "application/json"
A license is needed for the agent user to have a mailbox, Teams presence, etc. Usage location must be set first.
PATCH https://graph.microsoft.com/beta/users/{agent-user-id}
Content-Type: application/json
Authorization: Bearer <token>
{
"usageLocation": "US"
}
GET https://graph.microsoft.com/beta/subscribedSkus?$select=skuPartNumber,skuId,consumedUnits,prepaidUnits
Authorization: Bearer <token>
Requires Organization.Read.All permission.
POST https://graph.microsoft.com/beta/users/{agent-user-id}/assignLicense
Content-Type: application/json
Authorization: Bearer <token>
{
"addLicenses": [
{ "skuId": "<sku-id>" }
],
"removeLicenses": []
}
Connect-MgGraph -Scopes "User.ReadWrite.All","Organization.Read.All" -TenantId "<tenant>" -NoWelcome
# Set usage location
Invoke-MgGraphRequest -Method PATCH `
-Uri "https://graph.microsoft.com/beta/users/<agent-user-id>" `
-Body '{"usageLocation":"US"}' -ContentType "application/json"
# Assign license
$licenseBody = '{"addLicenses":[{"skuId":"<sku-id>"}],"removeLicenses":[]}'
Invoke-MgGraphRequest -Method POST `
-Uri "https://graph.microsoft.com/beta/users/<agent-user-id>/assignLicense" `
-Body $licenseBody -ContentType "application/json"
Tip: You can also assign licenses via the Entra admin center under Identity → Users → All users → select the agent user → Licenses and apps.
| Service | Estimated Time | |---|---| | Exchange mailbox | 5–30 minutes | | Teams availability | 15 min – 24 hours | | Org chart / People search | Up to 24–48 hours | | SharePoint / OneDrive | 5–30 minutes | | Global Address List | Up to 24 hours |
idtyp=user tokens)| Error | Cause | Fix |
|---|---|---|
| Agent user IdentityParent does not exist | identityParentId points to a non-existent or non-agent-identity object | Verify the ID is an agentIdentity service principal, not a regular app |
| 400 Bad Request (identityParentId already linked) | The agent identity already has an agent user | Each agent identity supports only one agent user |
| 409 Conflict on UPN | The userPrincipalName is already taken | Use a unique UPN |
| License assignment fails | Usage location not set | Set usageLocation before assigning licenses |
development
Generate breadboard circuit mockups and visual diagrams using HTML5 Canvas drawing techniques. Use when asked to create circuit layouts, visualize electronic component placements, draw breadboard diagrams, mockup 6502 builds, generate retro computer schematics, or design vintage electronics projects. Supports 555 timers, W65C02S microprocessors, 28C256 EEPROMs, W65C22 VIA chips, 7400-series logic gates, LEDs, resistors, capacitors, switches, buttons, crystals, and wires.
development
Apply lean thinking to UX: hypothesis-driven design, collaborative sketching, and rapid experiments instead of heavy deliverables. Use when the user mentions "Lean UX", "design hypothesis", "UX experiment", "collaborative design", or "outcome over output". Covers hypothesis statements, MVPs for UX, and cross-functional collaboration. For Build-Measure-Learn, see lean-startup. For usability audits, see ux-heuristics.
development
Design MVPs, validated learning experiments, and pivot-or-persevere decisions using Build-Measure-Learn. Use when the user mentions "MVP scope", "validated learning", "pivot or persevere", "vanity metrics", or "test assumptions". Covers innovation accounting and actionable metrics. For 5-day prototype testing, see design-sprint. For customer motivation analysis, see jobs-to-be-done.
tools
Instrument, trace, evaluate, and monitor LLM applications and AI agents with LangSmith. Use when setting up observability for LLM pipelines, running offline or online evaluations, managing prompts in the Prompt Hub, creating datasets for regression testing, or deploying agent servers. Triggers on: langsmith, langchain tracing, llm tracing, llm observability, llm evaluation, trace llm calls, @traceable, wrap_openai, langsmith evaluate, langsmith dataset, langsmith feedback, langsmith prompt hub, langsmith project, llm monitoring, llm debugging, llm quality, openevals, langsmith cli, langsmith experiment, annotate llm, llm judge.