UiPath Documentation
automation-ops
latest
false
Automation Ops user guide
  • Getting Started
    • Introduction
      • Governance
      • Feeds Management
      • Automation Ops™ User Roles
      • Licensing
    • Solutions Management
    • Automation Ops feature availability
  • Governance
  • Source Control
  • CI/CD Pipelines
  • Feed Management
  • Logging

Settings for Autopilot Policies

Settings for Autopilot governance policies in Automation Ops, covering available tools, MCP servers, security mode, custom instructions, models, and context grounding knowledge.

This page describes the settings available for Autopilot policies. To create a policy, refer to Create a governance policy. To apply it to users, groups, or tenants, refer to Deploy governance policies.

An Autopilot policy applies to Autopilot in every host: Studio, Studio Web, and Visual Studio Code. To turn Autopilot on or off, use Studio policies and Studio Web policies.

Policy details​

  • Policy Name - The name of the policy.
  • Description - A description of the policy.
  • Priority - Decides which policy applies when multiple group-level policies target the same user. A policy with a lower value applies before a policy with a higher value.

Tools​

On the Tools tab, enable or disable the individual tools that Autopilot can use. Tools are grouped by category:

CategoryTools for
File SystemReading, writing, and editing files, and listing directory contents
Shell / Script ExecutionRunning Bash and PowerShell commands
Search & WebSearching the web, reading web pages, and searching files by name or content
UtilityKeeping a to-do list for the current task

MCP servers​

On the MCP servers tab, control which Model Context Protocol (MCP) servers are available to Autopilot users.

  • Allow users to add their own MCP servers - When set to Yes, users can add personal MCP servers in Autopilot, under Settings > MCP servers. When set to No, only the organization MCP servers are available.
  • Organization MCP servers - Remote MCP servers provisioned for every user the policy applies to. Users connect to them automatically and cannot remove them. Each server has a Name and a Server URL.

Organization MCP servers must be remote endpoints reachable from the user's browser, using HTTP or Server-Sent Events (SSE) transport. To offer a different set of servers to a specific team, create a separate policy and deploy it to that user group or tenant.

Security​

On the Security tab, control how autonomously Autopilot operates.

  • Allow user to change security mode - When set to Yes, users can switch the security mode in Autopilot. When set to No, the mode selected in Security mode is enforced.
  • Security mode - Available when Allow user to change security mode is set to No. Available modes:
    • Cautious - Autopilot asks for approval before every operation and runs processes in strict isolation.
    • Adaptive - Autopilot runs safe tools automatically and asks for approval before risky operations.
    • Full access - Autopilot has full system access and does not ask for approval.
  • Custom instructions - Organization-level instructions added to the start of the system prompt in every Autopilot conversation. They set the response style, domain context, or organizational conventions. They are combined with the personal custom instructions of each user, and the organization instructions come first.

Model availability, personally identifiable information (PII) masking, guardrails, audit logging, and model region routing are set in the AI Trust Layer policy. For details, refer to Settings for AI Trust Layer policies.

Models​

On the Models tab, select the models available to Autopilot users and the default model for new conversations.

  • Default model - The model that new conversations use. Users can switch to another enabled model.
  • Available models - Enable or disable individual models, grouped by provider.
Important:

A model you enable on this tab is available in Autopilot only if its provider is also enabled on the Models tab of the AI Trust Layer policy, at tenant level. For details, refer to Models in Settings for AI Trust Layer policies.

Knowledge​

On the Knowledge tab, select Enable context grounding knowledge to let Autopilot use organization knowledge bases and document indexes for grounded responses.

  • Policy details​
  • Tools​
  • MCP servers​
  • Security​
  • Models​
  • Knowledge​

Was this page helpful?

Connect

Need help? Support(opens in new window)

Want to learn? UiPath Academy(opens in new window)

Have questions? UiPath Forum(opens in new window)

Stay updated