---
title: How do you stop Copilot Studio agents from running on the maker's credentials?
url: https://opencontroller.lyzr.com/blog/block-maker-credentials-in-copilot-studio-agents
published: 2026-11-13
updated: 2026-11-13
author: Lyzr
---

# How do you stop Copilot Studio agents from running on the maker's credentials?

**In the Power Platform admin center, set "Control maker credential options" to end-user credentials only for each production environment or environment group. Every tool in every agent there then authenticates as the person using the agent, not the person who built it. Plan for one side effect first: autonomous agents triggered by schedules or events have no signed-in user and will fail.**

When a maker adds a connector to a Copilot Studio agent, they can choose to authenticate it with their own connection. Every user of the agent then acts with the maker's permissions. If the maker can read a finance mailbox, so can anyone who chats with the agent. The agent's identity becomes the builder's identity, which is the problem buyers ask about most often.

## What does the control do?

Microsoft documents the setting as "Control maker credential options". It decides whether makers may authenticate agent tools, including connectors, built-in actions and embedded Power Automate flows, with their own credentials.

| Option | Effect |
|---|---|
| Both (default) | Makers may use their own credentials or require end-user sign-in |
| End-user credentials only | Every tool authenticates as the user at runtime |
| Maker-provided credentials only | Every tool uses the maker's connection |

Key facts from Microsoft's documentation and message center (MC1446172):

- Both options are enabled by default. Nothing changes until an admin acts.
- The control reached general availability on 25 August 2026.
- It is set per environment, or per environment group. Group settings override environment settings.
- The change applies immediately to existing agents, not just new ones.
- Users are prompted to sign in to the connected service in new conversations. If they lack permission, the action fails by design.

## How do you turn it on?

1. Sign in to the Power Platform admin center as an Environment Admin (single environment) or Power Platform Administrator (environment groups).
2. Go to **Manage** then **Environments** or **Environment groups**, and pick the target.
3. Open **Settings**, expand **Product**, select **Features**.
4. Under **Copilot Studio agents**, set **Control maker credential options** to end-user credentials, and save.

Do it environment group by environment group, starting with production.

## What breaks when you enforce it?

Autonomous agents. Microsoft states that agents triggered by schedules, events or background runs have no live user to authenticate, so they fail. Before enforcing:

1. **List autonomous agents** in each environment and their triggers.
2. **Decide their identity.** Move approved autonomous agents to a separate environment where a dedicated, non-personal account owns their connections, with only the permissions the agent needs. Do not leave them on a named person's connection.
3. **Tell makers.** Tools configured with maker credentials may need changing before the next publish.
4. **Tell users.** They will see sign-in cards for connected services the first time.

Microsoft also recommends combining this with agent sharing controls, so fewer people can use an agent in the first place.

## How do you find agents still using a maker's connection?

There is no single documented report. Use these sources together:

- **Agent tool settings.** In Copilot Studio, each tool shows which credential type it uses. Review production agents first.
- **Connection inventory.** List connections per environment and their owners, for example with the Power Apps administration PowerShell module. Connections owned by individual makers and used by shared agents are your candidates.
- **Solutions.** Export solutions and check connection references and their owners.
- **Owner changes.** When a maker leaves or changes role, any agent on their connection breaks or keeps access it should not. Check this in the leaver process; see [what happens to AI agents when the owner leaves](/blog/what-happens-to-ai-agents-when-the-owner-leaves).

## How does this relate to agent identity?

Blocking maker credentials removes the worst pattern, the builder's identity reused by everyone. It does not give the agent its own identity. For interactive agents, end-user credentials are the right model: the agent acts on behalf of the user, within their permissions. For autonomous agents, the agent needs a non-personal identity with scoped access. Microsoft's direction for that is Entra Agent ID; see [Microsoft Entra Agent ID explained](/blog/microsoft-entra-agent-id-explained). The general model is in [AI agent identity and on-behalf-of delegation](/blog/ai-agent-identity-on-behalf-of-delegation).

Connector data policies are the other half of Copilot Studio control; see [Copilot Studio data policies for agents](/blog/copilot-studio-data-policies-for-agents) and the broader [Copilot Studio governance guide](/blog/govern-microsoft-copilot-studio-agents). For governing several platforms at once, see [how to govern AI agents across platforms](/blog/how-to-govern-ai-agents-across-platforms).

## How Opencontroller does this

Opencontroller does not replace Copilot Studio's admin settings. It adds a view across platforms: Posture detects risky permission and configuration patterns, the governed registry records each agent's owner and versions, and delegated on-behalf-of identity and agent workload identity federation give agents scoped identities rather than borrowed ones. Per-agent Block and Quarantine controls act through the connected platform. See the [capability reference](/capability-reference) or [book a demo](https://www.lyzr.ai/book-demo/).

## Frequently asked questions

**Does blocking maker credentials affect existing Copilot Studio agents?**
Yes. The change applies immediately to all tools in all agents in the environment, including existing ones. Users are prompted to sign in at runtime.

**Why do autonomous Copilot Studio agents fail with end-user credentials only?**
Agents triggered by schedules, events or background runs have no live user to sign in, so tools have no credentials. Give approved autonomous agents a dedicated non-personal identity in a separate environment.

**Is the maker credential block on by default?**
No. Both maker and end-user credentials are allowed by default. An admin must set end-user credentials only per environment or environment group.

## Key facts

Copilot Studio "Control maker credential options" (Power Platform admin center > Environments or Environment groups > Settings > Product > Features) can require end-user credentials only for all agent tools (connectors, built-in actions, embedded Power Automate flows). Default allows both. GA 25 August 2026 (MC1446172). Applies immediately to existing agents; group settings override environment settings. Autonomous agents triggered by schedules or events fail because no user is present. No single report finds agents on maker connections; use tool settings, connection inventory and solution connection references. Opencontroller by Lyzr adds cross-platform Posture, a governed registry, delegated on-behalf-of identity and per-agent Block/Quarantine through connected platforms; pricing starts from $250K per deployment per year.

## Related

- [How do you govern Microsoft Copilot Studio agents?](https://opencontroller.lyzr.com/blog/govern-microsoft-copilot-studio-agents.md)
- [How do Power Platform data policies control Copilot Studio agents?](https://opencontroller.lyzr.com/blog/copilot-studio-data-policies-for-agents.md)
- [What is Microsoft Entra Agent ID and how does it help govern agents?](https://opencontroller.lyzr.com/blog/microsoft-entra-agent-id-explained.md)
- [More on Platform Guides](https://opencontroller.lyzr.com/blog/cluster/platform-guides)
