Agent
Using Tasklist
Use Tasklist to break down complex problems into manageable steps, track progress, and collaborate with Agent on multi-step tasks.
What is the Tasklist?
Augment’s Tasklist helps the Agent create and refine a step-by-step plan for you to review. The Tasklist provides a structured interface for collaboration between you and the Agent, allowing you to break down complex problems into manageable, sequential steps.
Why Tasklist?
Tasklist improves agent effectiveness on long or complex tasks by:- Maintaining context across different conversations by moving your Tasklist to a new chat
- Breaking down complex problems into manageable, sequential steps
- Gathering progress across threads
- Exploring alternative solutions to completed tasks if you need to pivot
- Streamlining your approach to nebulous problems by deleting irrelevant steps once the path forward is clear
Tasklist in Action
Creating a New Task
Automatic Creation
The Agent will usually create a Tasklist when it encounters a complex, multi-step problem. You can also ask the Agent to make a Tasklist for you by simply prompting “Start a Tasklist to…” then add the problem you are trying to tackle.Manual Creation
You can also manually create a Tasklist:- Switch to Tasklist using the checklist button next to Changes
- Click the plus to add your first task
- Alternatively, you can create a new task by typing in the gray prompt box at the bottom of the extension. Click Add Task from the dropdown arrow next to Send
Running Tasks
To run a task, click the grey triangle (play button) next to the task. The Agent will begin executing the task and update its status as it progresses.
Task Status Indicators
Task statuses are indicated by different colors and icons:- Empty circle - Task has not yet started
- Blue half circle - Task is currently in progress
- Green checkbox - Task has been completed and is ready for review
Subtasks
Augment Code automatically generates subtasks when needed. The Agent will automatically add and nest required subtasks under your initial tasks. You can edit and expand these subtasks just like any other task in the list. Likewise, you can remove subtasks you deem unnecessary.
Managing Running Tasks
Stopping a Task
You can treat any in-progress task like any prompt you might send the Agent. To stop what the Agent is doing and offer a corrective action, click the red square (stop button) and tell the Agent what you want it to do instead.Running All Tasks
The Agent can complete all the tasks sequentially by clicking the triangle (play button) at the top of the Tasklist.Reviewing Changes
You can review the changes made by the Agent after a task is completed by toggling between the Tasks and Changes view to see the diffs (differences) of the work done by the agent for each task.
Integration with Task Management Tools
Jira and Linear Integration
The Tasklist is a perfect pairing with existing task management tools like Jira or Linear:- Ask the Agent to create a Tasklist based on tickets inside Jira or Linear
- Further break down complex tickets into manageable steps
- Once your Tasklist is completed, you can ask the Agent to resolve the issue inside Jira or Linear and append the steps taken as a comment
Standalone Usage
Don’t use an issue tracker? No problem - use Tasklist to track issues you need to tackle across Threads.Best Practices
- Be specific when creating tasks to help the Agent understand exactly what needs to be done
- Review and edit the automatically generated subtasks to ensure they align with your goals
- Use the stop function to provide course corrections when the Agent is heading in the wrong direction
- Leverage the Changes view to review all modifications made during task execution
- Move Tasklists between conversations to maintain context across different chat sessions