Sharing

Skill-level access control allows administrators to define which users or groups can access and run specific skills within an agent. This enables more granular security and ensures that sensitive or specialized capabilities are only available to authorized users.

Overview

Each agent consists of one or more skills. By default, all skills within an agent are accessible to any user who has permission to use the agent. Skill-level access control provides the ability to override this default behavior by restricting access to individual skills.

Access to a skill is determined by two factors:

1. The user must have access to the agent.
2. The user must have explicit permission to the skill.

If a user has access to the agent but not to a restricted skill, the skill will not be visible or executable at runtime.

Configuring Access

Skill access can be configured directly within the agent:

1. Navigate to the agent that contains the desired skill.
2. Open the skill’s settings.
3. In the Sharing section, specify the users or groups that should have access to the skill.
4. Save the configuration.

Once saved, only the designated users or groups will be able to access and run the skill.

Runtime Behavior

At runtime, the set of available skills for a user is determined by their permissions. Users without access to a restricted skill will not see it in the agent’s available actions and will not be able to execute it. This enables a single agent to present different sets of capabilities depending on the user’s permissions.

Use Cases

• Security: Restrict access to skills that expose sensitive data or critical system actions.
• Role-Based Functionality: Allow different teams to use different capabilities within the same agent.
• Governance: Ensure that only approved users can access and run specific analytical or operational functions.

Summary

Skill-level access control provides fine-grained permission management within agents. By combining agent-level and skill-level access requirements, it ensures that users can only interact with the skills they are authorized to use, supporting both security and flexible user experiences.