Sometimes the best automations are the simplest. In this quick-start guide, you’ll build a Use Case that triggers when someone @mentions an Overcut agent in a ticket or issue. Once triggered, the workflow will clone the relevant repository and launch an agent session to handle the request autonomously.

​
Common use cases:

  • Drafting a spec
  • Writing documentation
  • Generating tests
  • Investigating bugs

​
Create a New Use Case

1

Open the Use Cases page

From the main menu, click Use Cases and then select Create Use Case.
2

Name your Use Case

Give your Use Case a descriptive name like Ticket Mention Responder.

​
Set up the Trigger

1

Edit the default trigger

When the Use Case is created, it includes a default trigger node. Click on the node to open the properties panel.
2

Select the Mention trigger

In the trigger type selector, choose Mention. This will activate the workflow whenever an agent is mentioned in a supported system (e.g. GitHub, GitLab, Bitbucket, Jira).
You can set additional conditions, such as limiting the trigger to specific repositories or users.
3

Note about execution timing

Mention triggers always execute immediately when detected, ensuring responsive user interaction. Delay settings are not applicable for this trigger type.

​
Add the Clone Repo Action

1

Add the action to the canvas

From the left sidebar, drag the Clone Repo action onto the canvas and connect it to the Mention trigger node.
2

Select the Git Organization

Choose your organization from the dropdown list of connected GitHub organizations.
3

Select the Repository Name

Enter the name of the repository to clone. For example: docs.
4

Select the Branch

Specify the branch to clone. For example: main.
You can use dynamic values like {{trigger.issue.title}} to pull context from the issue.
5

Set the Alias (optional)

Set a friendly alias for the repository, such as product-docs. This helps agents refer to it clearly during the session.

​
Add the Agent Action

You have two options for handling mentions, depending on the complexity of the request:

​
Option 1: Simple Response with agent.run

For straightforward requests that can be handled by a single agent:
1

Add the Agent Run block

Drag the agent.run action to the canvas and connect it to the Clone Repo action.
2

Select the Agent

Choose a single agent that can handle the request (e.g., Senior Developer for code tasks, Tech Writer for documentation).
3

Set the Instructions

Enter specific instructions like: Analyze the user’s request and provide a helpful response. If they need documentation, create it. If they need code analysis, provide insights.

​
Option 2: Complex Interaction with agent.session

For complex requests that require multiple agents or ongoing conversation:
1

Add the Agent Session block

Drag the agent.session action to the canvas and connect it to the Clone Repo action.
2

Select Participating Agents

Choose the agents that should participate in the session. You can select multiple roles to cover different types of tasks.
You can define agent personas with specific skills on the Agents page.
3

Define the Session Goal

Enter the goal of the session. Example: Help the user with their request.
This serves as the top-level instruction for the coordinator agent.
4

Set the Instructions

Enter a prompt like: Follow the instructions from the user on the issue in context.
This is a generic instruction. For more advanced use cases, you can provide detailed, structured prompts based on issue metadata, labels, or ticket content.
When to use each action type? Use agent.run for simple, one-time responses. Use agent.session for complex requests that need multiple agents or ongoing conversation. Learn more in our Agent Differences guide.

​
Test the Workflow

1

Mention the Agent in a Ticket

Open an issue or ticket and mention your agent with a request, for example:
@overcut Can you write the README for this branch?
2

Check the Execution

After triggering, you should see a new entry in the Executions page. This confirms the workflow has been activated.
3

Verify the Response

Check that the agent responded appropriately:
  • For agent.run: Single response with the requested information
  • For agent.session: Ongoing conversation that can handle follow-up requests
The agent will post its results or replies directly into the original ticket or thread. With agent.session, the conversation continues as you add more comments.

​
Expand the Flow

1

Respond with follow-up comments

Try posting additional comments to the issue. Overcut will keep listening and respond within the same context as part of the running session.
2

Add logic or notifications (optional)

You can enhance the workflow by adding conditional logic, Slack alerts, or branching behaviors based on content or label detection.

​
✅ Success Checklist

  1. Workflow triggers when an agent is mentioned in a ticket
  2. The correct repository and branch are cloned
  3. The agent responds appropriately to the request
  4. The agent posts results directly in the ticket thread

​
Next Steps

Now that you’ve built a mention response workflow, explore these related topics: