Build a feature with Agent mode
Goal. Hand a whole feature to Agent mode and let it write the multi-file changes for you.
I have a CSV called expression_data.csv with gene names in column A and values for 6 samples in B–G. Write a Python script that loads it with pandas, drops any gene with >20% missing values, and plots a clustered heatmap with seaborn.
Switch the chat to Agent mode and send this. The agent reads your project, writes the files, and plans the steps itself — you review the proposed changes before they apply.
- 1Turn on Agent mode in the chat sidebar. Unlike a single chat answer, the agent autonomously reads the project and writes changes across multiple files.
- 2Send the prompt above. The agent generates a complete, ready-to-run script — imports, the >20%-missing cleaning logic, and a seaborn clustermap — and shows you the diff.
- 3Keep the ask specific and small-scoped: name the file, the columns, and the exact rule. Vague instructions let the agent drift; precise ones keep it on track.
You'll see. A finished, runnable script (imports, data cleaning, and a clustered heatmap) written across the files it needs — from one plain-English description.
Cost. Agent runs consume included usage by request; Hobby has limited Agent requests, so spend them on real features. Pro includes generous usage for Agent and Composer.
Takeaway. Agent mode turns a plain-English feature request into multi-file code — keep the scope tight for the best results.