You are reading the built-in manual
These are the pages AURA carries within itself: inside the application, the ? key opens the one for the screen you are on. They are published here exactly as they are, so they can be read without installing anything.
The inventory of projects Claude Code has already worked on. Each row is a folder under ~/.claude/projects, whose name encodes the source path.
Where the columns come from
| Column | Source |
|---|---|
| Project | The last segment of the real source path; the slug otherwise |
| Path | The real working folder path, with Windows separators |
| Sessions | The number of .jsonl files at the root of the folder |
| Size | The total weight of the transcript folder, sub-agents included |
| Last activity | The modification date of the most recent transcript |
The real path is not derived from the slug: the slug replaces every non-alphanumeric character with a dash, which cannot be undone. I recover it by reading the paths declared in ~/.claude.json and re-encoding them to find a match. A project that file says nothing about therefore shows its raw slug instead of a path — a sign that no match could be established, not that the project is broken.
The badge next to the name means the source folder holds a .claude folder — so, resources of its own.
The default sort puts the last activity first. Every column sorts, and the filter covers both the name and the path.
How to read this list
- Last activity tells a living project from an abandoned one.
- Size says which one weighs, before going to Maintenance.
- The session count counts transcripts, not distinct conversations in the human sense: a session resumed later stays one file.
Clicking a row — anywhere on it — opens the project detail.
Orphan projects
A project whose sources were moved or deleted keeps its transcripts: they stay listed, readable and replayable. Only its .claude resources become unreachable, since they lived in the source tree.