Skip to main content

​
Overview

Basic Memory Cloud uses rclone bisync for bidirectional file synchronization between your local filesystem and cloud storage. This enables:
  • Two-way sync: changes in either location are propagated to the other
  • Conflict detection and resolution
  • Integrity verification with checksums
  • Incremental sync (only changed files)
  • Dry-run preview before syncing
Cloud sync requires an active subscription and authenticated cloud access. Run bm cloud login before setting up sync.

​
Initial Setup

​
Step 1: Install and Configure rclone

Run the cloud setup wizard to install rclone and configure credentials: 
bm cloud setup
This command:
  1. Downloads and installs rclone (if not present)
  2. Fetches your tenant information from the cloud API
  3. Generates scoped S3 credentials for sync
  4. Configures the rclone remote named basicmemory-cloud
Example output: 
Basic Memory Cloud Setup
Setting up cloud sync with rclone...

Step 1: Installing rclone...
✓ rclone already installed at /usr/local/bin/rclone

Step 2: Getting tenant information...
✓ Found tenant: 11111111-1111-1111-1111-111111111111

Step 3: Generating sync credentials...
✓ Generated secure credentials

Step 4: Configuring rclone remote...
✓ Remote 'basicmemory-cloud' configured

✓ Cloud setup completed successfully!

Next steps:
1. Add a project with local sync path:
   bm project add research --cloud --local-path ~/Documents/research

   Or configure sync for an existing project:
   bm project sync-setup research ~/Documents/research

2. Preview the initial sync (recommended):
   bm cloud bisync --name research --resync --dry-run

3. If all looks good, run the actual sync:
   bm cloud bisync --name research --resync

4. Subsequent syncs (no --resync needed):
   bm cloud bisync --name research

Tip: Always use --dry-run first to preview changes before syncing

​
Step 2: Configure Project Sync

Each project needs a local sync path configured:

​
Option A: Create New Project with Sync

bm project add research --cloud --local-path ~/Documents/research

​
Option B: Configure Existing Project

bm cloud sync-setup research ~/Documents/research
This command:
  • Verifies the project exists in the cloud
  • Creates the local directory if needed
  • Updates project configuration with sync path
  • Creates the project in local database for MCP access
The local sync path should be a dedicated directory for the project. Don’t use your default Basic Memory directory (~/basic-memory) unless that’s specifically what you want to sync.

​
Sync Operations

​
Bidirectional Sync (bisync)

The primary sync operation for two-way synchronization: 
# First-time sync (establishes baseline)
bm cloud bisync --name research --resync --dry-run  # Preview
bm cloud bisync --name research --resync             # Perform

# Subsequent syncs (incremental)
bm cloud bisync --name research --dry-run  # Preview
bm cloud bisync --name research             # Perform

​
Command Options

OptionDescription
--nameProject name to sync (required)
--resyncForce new baseline (first sync or after conflicts)
--dry-runPreview changes without applying them
--verboseShow detailed output including file-by-file operations

​
First Sync: Resync Required

The first sync for a project requires --resync to establish a baseline: 
bm cloud bisync --name research --resync
What happens:
  1. rclone scans both local and cloud directories
  2. Creates a baseline snapshot of all files
  3. Stores metadata in ~/.cache/basic-memory/bisync/research/
  4. Synchronizes any differences (usually cloud → local on first sync)
  5. Marks the project as initialized in config
Always use --dry-run on first sync to preview what will be transferred:
bm cloud bisync --name research --resync --dry-run
Review the output carefully before running without --dry-run.

​
Incremental Syncs

After the initial resync, subsequent syncs are incremental: 
bm cloud bisync --name research
What happens:
  1. rclone compares current state against baseline
  2. Identifies changes in both local and cloud
  3. Propagates changes bidirectionally
  4. Detects and reports conflicts
  5. Updates baseline with new state

​
One-Way Sync (local → cloud)

For scenarios where you want to force the cloud to match local: 
# Preview
bm cloud sync --name research --dry-run

# Perform
bm cloud sync --name research
One-way sync (cloud sync) makes the cloud identical to local, potentially deleting cloud files not present locally. Use with caution.

​
Integrity Check

Verify file integrity between local and cloud without syncing: 
# Full check (both directions)
bm cloud check --name research

# One-way check (faster)
bm cloud check --name research --one-way
Example output: 
Checking research integrity...
✓ research files match
Or if differences exist: 
Checking research integrity...
⚠ research has differences

​
Conflict Resolution

​
Understanding Conflicts

A conflict occurs when the same file is modified in both locations between syncs: 
Error: bisync critical error: path1 and path2 are out of sync
Please run with --resync to establish a new baseline

​
Resolving Conflicts

Step 1: Review Changes Check what changed locally and in the cloud before resolving. Step 2: Backup Important Changes If you have important local changes, back them up: 
cp -r ~/Documents/research ~/Documents/research-backup
Step 3: Decide on Resolution Strategy

​
Strategy A: Cloud Wins (Download latest)

# Preview what will be downloaded
bm cloud bisync --name research --resync --dry-run

# Apply cloud changes
bm cloud bisync --name research --resync

​
Strategy B: Local Wins (Upload latest)

# Make cloud match local exactly
bm cloud sync --name research --dry-run
bm cloud sync --name research

# Re-establish bisync baseline
bm cloud bisync --name research --resync

​
Strategy C: Manual Merge

  1. Manually reconcile conflicting files
  2. Run resync to establish new baseline: 
    bm cloud bisync --name research --resync

​
Preventing Conflicts

  • Sync frequently: Run bisync regularly on each device
  • Sync before major changes: Always sync before editing many files
  • Use snapshots: Create snapshots before risky operations
  • Avoid simultaneous edits: Don’t edit the same files on multiple devices simultaneously

​
Sync State Management

​
Bisync State Files

Bisync stores state in ~/.cache/basic-memory/bisync/<project>/: 
~/.cache/basic-memory/bisync/research/
├── basicmemory-cloud-research.path1.lst-new
├── basicmemory-cloud-research.path1.lst-old
├── basicmemory-cloud-research.path2.lst-new
└── basicmemory-cloud-research.path2.lst-old
These files track the baseline state for incremental sync.

​
Resetting Bisync State

If bisync gets into an inconsistent state, reset it: 
bm cloud bisync-reset research
Output: 
✓ Cleared bisync state for project 'research'

Next steps:
  1. Preview: bm cloud bisync --name research --resync --dry-run
  2. Sync: bm cloud bisync --name research --resync
This removes state files, forcing a fresh --resync on next sync.

​
When to Reset State

  • Persistent conflict errors that won’t resolve
  • After manually modifying many files on both sides
  • When cloud path changes (different workspace/bucket)
  • After restoring from snapshot

​
Ignore Patterns

​
.bmignore File

Basic Memory uses .bmignore (gitignore-style syntax) to exclude files from sync: Location: ~/.basic-memory/.bmignore Default patterns: 
# Version control
.git
.svn

# Dependencies
node_modules
venv
.venv

# Build outputs
dist
build
target

# IDE
.vscode
.idea
*.swp
*.swo

# OS
.DS_Store
Thumbs.db

# Temporary
*.tmp
*.temp
.cache

​
Editing Ignore Patterns

  1. Edit ~/.basic-memory/.bmignore with your preferred editor
  2. Add patterns using gitignore syntax: 
    # Ignore all Python cache files
__pycache__
*.pyc

# Ignore specific directory
secrets/

# Ignore file type
*.log
  3. The patterns automatically convert to rclone filter format on next sync
The .bmignore.rclone file is auto-generated from .bmignore. Edit .bmignore, not the .rclone file.

​
Database Synchronization

After file sync completes, Basic Memory triggers a database sync: 
bm cloud bisync --name research
What happens:
  1. Files sync via rclone bisync
  2. CLI calls cloud API: POST /api/projects/{id}/sync
  3. Cloud server indexes new/changed files into SQLite database
  4. Knowledge graph updates with new entities and relations
Output: 
Bisync research (local <-> cloud)...
✓ research bisync completed successfully
Database sync initiated: Sync task started
Database sync happens automatically after successful file sync. You don’t need to trigger it manually.

​
Sync Workflows

​
Daily Sync Routine

# Morning: sync before starting work
bm cloud bisync --name research

# Work on files locally...

# Evening: sync before logging off
bm cloud bisync --name research

​
Multi-Device Workflow

On Device A: 
# Sync at end of work session
bm cloud bisync --name research
On Device B: 
# Sync before starting work session
bm cloud bisync --name research

# Work on files...

# Sync when done
bm cloud bisync --name research

​
Pre-Deployment Sync

Before deploying changes or major edits: 
# Create snapshot for safety
bm cloud snapshot create "before major refactor"

# Ensure everything is synced
bm cloud bisync --name research

# Verify integrity
bm cloud check --name research

​
Troubleshooting

​
Sync Fails: rclone Not Found

# Run setup again to install rclone
bm cloud setup

​
Sync Fails: Credentials Invalid

# Check cloud auth status
bm cloud status

# Re-login if needed
bm cloud login

# Regenerate sync credentials
bm cloud setup

​
Sync Fails: Permission Denied

# Check local directory permissions
ls -la ~/Documents/research

# Ensure you own the directory
sudo chown -R $USER ~/Documents/research

​
Sync Fails: Path Not Found

# Verify project sync configuration
bm project list

# Reconfigure sync path if needed
bm cloud sync-setup research ~/Documents/research

​
Files Not Syncing

  1. Check if files are ignored: 
    cat ~/.basic-memory/.bmignore
  2. Run sync with verbose output: 
    bm cloud bisync --name research --verbose
  3. Verify files exist locally: 
    ls -la ~/Documents/research

​
Persistent Conflicts

If conflicts won’t resolve: 
# Reset bisync state
bm cloud bisync-reset research

# Review what will change
bm cloud bisync --name research --resync --dry-run

# Establish fresh baseline
bm cloud bisync --name research --resync

​
Performance Optimization

​
Large Projects

For projects with many files:
  • Use ignore patterns to exclude unnecessary files
  • Sync regularly for smaller incremental updates
  • Enable verbose output only when debugging
  • Consider multiple smaller projects instead of one giant project

​
Network Optimization

  • Sync during off-peak hours for large transfers
  • Use wired connection for initial resync
  • Monitor bandwidth with --verbose flag

​
Next Steps

Project Management

Advanced cloud project configuration

Snapshots

Create and restore from snapshots

Authentication

Manage cloud credentials

Build docs developers (and LLMs) love

Get started for freeTalk to us