Lock agent

Overview

Lock agent is a protection feature that freezes an agent's configuration to prevent accidental or unauthorized modifications. Once locked, all agent settings—including instructions, AI models, skills, knowledge base sources, automations, and deployment options—are protected from edits.

This feature ensures stability for agents deployed in production or shared across teams. While configuration editing is restricted, authorized team members can continue to interact with and use the agent in chat without interruption.

Prerequisites

Before locking or unlocking an agent, ensure you meet the following requirements:

• Role permissions: You must be an Owner or Admin of the agent (or have workspace administrator privileges). Standard users cannot lock or unlock agents.

• Access: You must have access to the agent's setup page in Swiftask.

Impact on users without edit permissions:

• They can continue to chat with the agent normally.

• They cannot edit any settings, instructions, skills, or data connections.

• When opening the agent setup, they see a message stating that the agent is locked.

Step-by-step guide

1. Open the agent

Navigate to Agents in the left sidebar and select the agent you want to manage.

2. Lock the agent

In the top-right navigation bar, click the Lock agent button located next to Share and Try it.

3. Verify the locked status

Once locked, the button text changes to Unlock agent. An informational banner appears at the top of the agent settings: "This agent is locked. Please unlock it to perform changes."

All configuration fields are now read-only to prevent changes.

4. Unlock the agent to make changes

To edit the agent configuration, an Owner or Admin must click Unlock agent in the top-right bar. The banner disappears and all settings become editable again.

What is protected when locked?

When an agent is locked, the following configuration areas are protected:

• Profile: Name, description, department, avatar, and interface layout.

• Instructions & LLM: System prompts, model selection, temperature settings, and prompt versions.

• Skills & Knowledge base: Adding, removing, or editing tools, integrations, and connected documents.

• Automations: Scheduled tasks and event triggers.

• Deployment: Chat widget customizations, Agent as app options, and API configurations.

• Agent management: Duplicating, deleting, or transferring ownership of the agent.

Practical use cases

• Production agents: Secure customer-facing or mission-critical agents against inadvertent edits to prompts or knowledge bases.

• Shared corporate templates: Protect standard internal agents shared across teams so that core instructions remain consistent.

• Ownership handovers: Keep an agent locked while transitioning management responsibilities to a new team member.

Tips & best practices

• Lock after major releases: Once you test and validate prompt updates or tool integrations, lock the agent immediately to preserve stability.

• Keep team members informed: Inform your team that the agent is locked so they can submit configuration change requests to designated Admins or Owners.

• Unlock only during active maintenance: Temporarily unlock the agent while making verified updates, then re-lock it as soon as changes are applied.

Troubleshooting

Additional resources

• Share an agent – Manage team member access and assign Owner, Admin, or User roles.

• Create an AI agent step by step – Learn how to build and configure custom AI agents.

• Embed and widget – Set up and embed your agents on external websites or apps.

• Governance and workspace administration – Configure workspace roles and organization permissions.