refactor(strategy): sweep push-bootstrap marker on disable, not via entire clean

main

Commit

karthik-rameshkumar2mo ago

Follow-up to the review fix: the earlier commit enrolled the empty-remote bootstrap marker in entire clean's typed cleanup machinery, but that command is scoped to session data (session states, shadow branches, temp files), and the marker is a per-remote performance cache — not session data.

The sibling entire-session-locks directory (also a non-session-data cache) establishes the right pattern: it is swept by entire disable's pristine teardown via a best-effort os.RemoveAll, and is not an entire clean item. Mirror that: drop the CleanupTypePushBootstrap enum, result fields, discovery, deletion, and clean-command display plumbing, and instead RemoveAll the marker dir in removeAllSessionStates alongside entire-session-locks. This fixes the real gap — the marker previously survived entire disable as leaked state — while removing the exhaustive-switch maintenance surface. The marker still self-heals via its TTL + fingerprint, so entire clean has no reason to touch it.

Co-Authored-By: Claude Opus 4.8 (1M context) noreply@anthropic.com Entire-Checkpoint: 01KXG26CGYEFX9015W2KTZ29FK

Checkpoints

Fix Checkpoint Push on Empty Remote

Claude CodeOpus 4.8[1m]
View session
Checkpoint 1