The PlainID MCP Server exposes PlainID capabilities to LLMs and AI agents over the Model Context Protocol. This allows administrators, developers, and security teams to query and manage authorization in natural language, while every action remains scoped to the permissions users already hold in the Tenant.
The server authenticates via OAuth, using the user's own PlainID credentials. There is no service account and no elevated bridge identity. So, a user with view access can read, a user with admin access can write - only within the Environments and Workspaces already granted to them.
Authentication is OAuth. On first connection, you will be prompted to sign in to your tenant with your own username and password. Everything you can subsequently do reflects the permissions already granted to you.
Use it to gain insight into your Tenant, analyze access paths, review recent activity, author Policies, and more.
How to use the PlainID MCP Server
There is no fixed set of commands. Instead, you ask in your own words. The areas below are common starting points, and the queries within them are illustrative rather than exhaustive. In practice, queries can cover any Policy, Access, or audit data within your permission scope. The following are example topics of how to gain insights or automate operations:
Understanding what's in your Tenant
Answers the query “What do we actually have, and where", which is useful for release reviews, audits, and tracking what is still in development versus what is approved and live in production.
Potential Queries
- "List all active Policies in the Production environment."
- "Compare the Policies in Dev/PolicyWS against Prod/PolicyWS and show me the difference."
- "Which Dynamic Groups appear most often across Policies in this Workspace?"
- "Show me every Policy attached to the payments application."
Understanding who can access what
Answers effective-access questions directly, rather than requiring you to infer them from Policy text, which are the questions auditors and incident responders tend to ask first.
Potential Queries
- "What can the invoice-processing agent access in this Environment?"
- "On whose behalf can the invoice-processing agent request access?"
- "Who can read customer financial records?"
- "Can the reporting service reach the loans dataset?"
- "Show me the full access topology for Production."
Creating Policies and Building Blocks
Describe the intent in plain language and have the authorization artifact drafted and applied — covering both new policies and amendments to existing ones, plus the Asset Templates and Applications they depend on.
Potential Queries
- "Create a Policy allowing regional managers to approve loans under $50,000."
- "Define an Asset Template for Bank Accounts with View, Approve and Close actions."
- "Set up a new Application for the claims portal using the existing claims template."
Understanding Recent Changes
Answers "What moved, when, and who moved it" — for change review, drift detection, and post-incident reconstruction.
Potential Queries
- "Summarize the changes made in this Environment over the last two weeks."
- "Who modified Policies in Production since Monday?"
- "Show me every Policy deleted this quarter."
- "Did any high-risk operations occur in the last month?"
Getting Started
Prerequisites
- An active PlainID Tenant. Your Tenant name is the one in your PlainID Tenant URL.
- PlainID user credentials. Your existing permissions apply: View for all insight capabilities, Admin for generate operations, which are evaluated per Environment and Workspace.
- An MCP-capable client.
Setup
Connect the PlainID MCP Server to your client using the URL matching your tenant:
https://mcp-oauth.<region>.plainid.io/<tenant-name>/mcp
Valid <region> values:
| Value | Use |
|---|---|
us1 |
Production — US |
eu1 |
Production — Europe, including UK tenants |
preprod |
Pre-production |
The CA region is not currently supported. Contact PlainID support for details.
The PlainID MCP Server lets you explore Policies, trace access paths, draft new Policies, and review recent changes through natural conversation, all while enforcing the same Tenant permissions you already hold.
If a query hits an unexpected permission boundary, check your Environment and Workspace access with your PlainID administrator, or contact PlainID support for setup help.