Cloud sync/bisync commands hardcode force_full=True instead of using incremental sync

[phernandez]

Problem

The bm cloud sync and bm cloud bisync CLI commands hardcode force_full=True when triggering the database sync after file sync. This causes every sync operation to bypass incremental change detection and do a full re-index, which is wasteful and slow.

This was a development-time convenience that was never reverted.

Logfire trace showing the issue:
https://logfire-us.pydantic.dev/basic-memory/basic-memory-cloud?q=trace_id%3D%27019d49615caaa705493e2a4f7cb762d6%27+and+span_id%3D%279f6b06b8360d8679%27&spanId=9f6b06b8360d8679&traceId=019d49615caaa705493e2a4f7cb762d6

The trace shows POST /proxy/my-vault/project/sync ? force_full='true' — the caller is sending force_full=true as a query parameter on every sync.

Location

src/basic_memory/cli/commands/cloud/project_sync.py

Two occurrences:

1. sync_project_command (line ~134):

return await ProjectClient(client).sync(
    project_data.external_id, force_full=True  # should be False
)

2. bisync_project_command (line ~209):

return await ProjectClient(client).sync(
    project_data.external_id, force_full=True  # should be False
)

Fix

Change both calls to force_full=False (or just remove the parameter since the default is already False).

The force_full=True behavior should only be used by explicit admin operations like reindex_all_tenants and bm doctor, not normal sync flows.

[claude]

Claude finished @phernandez's task in 1m 21s —— View job • claude/issue-706-20260401-1417 • Create PR ➔

---

Triage & Fix

Classification: Bug | Priority: High | Component: Cloud/CLI | Complexity: Simple

• Analyze issue
• Read source file and verify the bug
• Apply fix (change force_full=True → force_full=False in both locations)
• Add bug label
• Commit and push changes

What was fixed:

In src/basic_memory/cli/commands/cloud/project_sync.py, two calls to ProjectClient.sync() hardcoded force_full=True:

• Line 134 (sync_project_command)
• Line 209 (bisync_project_command)

Both were changed to force_full=False, so normal sync/bisync operations now use incremental change detection instead of triggering a full re-index every time.