Problem
After aggressive cleanup, operators still see hundreds of GiB used at the filesystem level, but DMC only reports its managed worktree footprint. The gap forces manual df / directory spelunking and makes cleanup feel incomplete even when DMC-safe cleanup is exhausted.
Desired behavior
Add a managed diagnostic report, e.g. workspace cleanup diagnose-disk or equivalent, that attributes disk usage across:
- DMC primaries
- DMC worktrees
- known reconstructable artifacts
- blocked cleanup candidates
- non-DMC large directories under the workspace root
- site/runtime plugin copies when relevant
Acceptance criteria
- Report total filesystem usage context and DMC-managed usage side-by-side.
- Sort largest buckets first.
- Include actionable next commands for safe cleanup lanes.
- Avoid raw destructive operations; diagnostic is read-only.
Context
This came from a cleanup run where DMC artifact cleanup reached zero candidates, but the volume still had ~500 GiB used and DMC-sized worktrees explained only ~135 GiB.
Problem
After aggressive cleanup, operators still see hundreds of GiB used at the filesystem level, but DMC only reports its managed worktree footprint. The gap forces manual
df/ directory spelunking and makes cleanup feel incomplete even when DMC-safe cleanup is exhausted.Desired behavior
Add a managed diagnostic report, e.g.
workspace cleanup diagnose-diskor equivalent, that attributes disk usage across:Acceptance criteria
Context
This came from a cleanup run where DMC artifact cleanup reached zero candidates, but the volume still had ~500 GiB used and DMC-sized worktrees explained only ~135 GiB.