Chaining tasks

Chaining tasks is a powerful feature for building up relevant context to the agent for solving the next task. By using this technique, you can have the agent map out how a feature works in your app, using the task as a short-term memory for relevant context, making it easier for the agent to make precise changes to your app even when it grows in size. This is better exemplified than described.

Debug an issue

First prompt

Describe the issue you experience, and tell the agent to create a task for investigating it and writing down findings as it goes.

Prompt

When I upload a PDF, I see the PDF in my file list,
but when I click Download nothing happens.

Create a task for debugging this and save your findings as comments
on the task as you go.

Second prompt

Provide the agent with context from the first prompt and start a task based on it

Prompt

Read task X and create a new task for solving this issue. Make sure you stick to the
problem at hand. I'd like to start working on this task right away.

Make changes to existing feature

First prompt

Tell the agent what you want to change with an existing feature and have the agent map out the required changes for this.

Prompt

I want to be able to upload multiple files at the same time from my AttachmentPage.

Create a task for mapping out how the existing feature work and what changes must
be made for this to support multiple files in one upload

Second prompt

Prompt

Read task X and create a task for solving this issue. Make sure you stick to the
problem at hand. I'd like to start working on this task right away.