# Provision Microsoft Entra Agent Identities

Agent identity provisioning and OAuth token exchange require coordinated Microsoft Graph configuration. This skill provides workflows for Blueprints, per-instance identities, permissions, and sidecar deployment.

## Install

```bash
npx skillstore add microsoft/entra-agent-id
```

## Metadata

- Status: approved
- Slug: microsoft-entra-agent-id
- Version: 1.0.1
- Author version: 1.0.1
- Skillstore revision: r2
- Version status: valid
- Tree hash: 051feea99719e293c0798e8a3646691656394ca2081e847982e1ec47adaeaab1
- Author: microsoft
- GitHub username: microsoft
- License: MIT
- Repository: https://github.com/microsoft/azure-skills/tree/main/.github/plugins/azure-skills/skills/entra-agent-id/
- Ref: 9e952417e76879bc9d853e1b8b2cd6d6d8d4a1c2
- Supported tools: Claude, Codex, Claude Code
- Audit status: complete
- Agent install advisory: confirmation\_required
- Manual install advisory: allowed
- Artifact signature: available
- Audit attestation: unavailable
- Human verification: not\_verified
- Risk factors: external\_commands, filesystem, network, env\_access
- Quality score: 38
- Quality tier: warning
- Public page: https://skillstore.pages.dev/skills/microsoft-entra-agent-id
- Manifest: https://skillstore.pages.dev/api/skills/microsoft-entra-agent-id/manifest

## Capabilities

- Creates Agent Identity Blueprints, BlueprintPrincipals, and per-instance Agent Identities through Microsoft Graph.
- Explains application and delegated permission grants for individual Agent Identities.
- Builds two-step fmi\_path token exchanges for autonomous and on-behalf-of flows.
- Configures managed identity federation and development client secrets on Blueprints.
- Documents cross-tenant token exchange requirements and common Entra error codes.
- Provides Docker, Kubernetes, Python, and TypeScript patterns for the AgentID sidecar.

## Use Cases

- Provision an agent identity hierarchy: Create a Blueprint, its principal, and separately auditable identities for each deployed agent instance.
- Implement runtime token exchange: Configure autonomous or on-behalf-of authentication with per-agent Microsoft Graph permissions.
- Deploy a polyglot authentication sidecar: Prepare local Docker or Kubernetes sidecar configuration for agents written in any supported language.

## Prompt Templates

### Check prerequisites

```
List the Entra roles, Graph permissions, tools, and tenant details required to create an Agent Identity Blueprint. Do not make changes.
```

### Plan identity provisioning

```
Create a step-by-step plan for a Blueprint named {name} with {count} Agent Identities. Include sponsors, permissions, validation, and rollback.
```

### Design an OBO flow

```
Design an on-behalf-of flow for {agent} using {scopes}. Explain Blueprint API configuration, per-agent grants, token audiences, and validation.
```

### Review a production deployment

```
Review my AgentID sidecar and cross-tenant token design. Identify excessive permissions, exposed endpoints, credential risks, tenant mistakes, and concrete remediations.
```

## Limitations

- Requires appropriate Microsoft Entra roles, Graph permissions, and administrator consent.
- Microsoft Graph Agent ID APIs can change during preview and require documentation verification.
- Client secrets are suitable only for local development; production requires federated or managed identity.
- The skill does not replace standard app registration, Azure RBAC, or agent authoring tools.

## Best Practices

- Verify current Microsoft documentation before applying Graph request bodies or SDK configuration.
- Use workload identity federation in production and keep sidecar endpoints private.
- Grant the minimum permissions to each Agent Identity and confirm propagation before testing.

## Anti Patterns

- Do not attach credentials directly to Agent Identity service principals.
- Do not expose AgentID sidecar token endpoints through public ports, ingress, or load balancers.
- Do not grant broad Microsoft Graph scopes when a smaller per-agent permission set works.

## Security Audit

- Audited at: 2026-07-24T03:32:58.95\+00:00
- Summary: Most static alerts are false positives caused by Markdown syntax, OAuth scope strings, official Microsoft endpoints, and bounded configuration examples. One example can forward an agent bearer token to an unrestricted URL, while semantic review found an exposed sidecar port and broad long-lived delegated grants.

## Stats

- Views: 87
- Downloads: 5
- Favorites: 0
- Popularity score: 0