Editor and files

Cone keeps the project beside the conversation, so you can read what the agent wrote without leaving the thread.

File tree and editor

The right rail shows the project’s file tree. Clicking a file opens it in a tabbed center editor with syntax highlighting, optional line wrapping, and line numbers. ⌘P opens a fuzzy file picker over the thread.

Code comments

Select lines in the editor or in a diff and attach a comment to the chat. The comment rides along with your next prompt as a chip carrying the file, the selected region, and your note, which beats pasting code into the composer by hand. Quoting a selection and asking about a whole file work the same way.

Terminal

Every thread has a persistent terminal in the right rail, running in the thread’s working directory. On remote threads it opens on the remote host over the same SSH connection the agent uses, which also makes it the place to complete interactive SSH authentication once.

Composer

• Mention files with @ to autocomplete paths from the project and attach them as context.
• Slash commands from the provider’s CLI autocomplete with /.
• Attach images by pasting or dragging them in.
• Edit an earlier prompt and resend it to branch the conversation from that point.

Transcript

Assistant output renders as full markdown, including tables, nested lists, and code fences. Consecutive tool calls collapse into task groups you can expand, thinking blocks fold away when finished, and a status line tracks what the agent is doing during a live turn. Turns that leave a background process watching (a CI run, a dev server) show it on the thread until it ends.