Changelog

⚡ Change Data Capture (CDC)

Real-time replication for PostgreSQL (logical WAL) and MySQL (binary log). Stream INSERT/UPDATE/DELETE events into your destination as they happen — no polling, no missed rows, sub-100ms lag. CDC checkpoints (LSN / binlog position) are persisted in the Context Engine so streams resume cleanly after a restart.

🛡️ Per-Record Fault Tolerance

A bad row no longer kills the pipeline. Failing rows are isolated, the rest of the batch commits, and every failure is recorded with its error type. Each sync response now reports Rows synced / Rows failed with a breakdown by type (null_violation, duplicate_key, type_mismatch…).

📋 Recipe Manager

Save any transform configuration as a named recipe. Apply it with one command across any environment. Frequently-used transforms are auto-captured from sync run history.

💡 Proactive Suggestions

Schema changes, recurring errors, and slow syncs are detected, scored, and surfaced via get_suggestions. Accept or dismiss with one command — dismissed suggestions are never re-raised. The Error Learner detects recurring failure patterns and links them to suggested recipes.

🔗 Dependency-Aware Scheduling

Express job dependencies as a DAG. Cycles are rejected at definition time. Failed parents block children and retry with exponential backoff (1m → 2m → 4m → 8m). scheduler_show_graph renders the full topology.

Tool Count

Upgrading from v2.0

🗄️ Three New Database Connectors

Datavor now connects to SQL Server, SQLite, and Snowflake — in addition to MySQL and PostgreSQL. Sync between any combination of all five engines in any direction.

Also compatible with cloud-hosted databases: Supabase, AWS RDS, Azure SQL, Aiven, PlanetScale — all accessed through the same MySQL/PostgreSQL/SQL Server drivers.

🧠 Data Context Engine

The most important feature in v2.0. Datavor builds a persistent local knowledge brain at ~/.datavor/context.db that silently learns from every interaction.

🔗 Universal Type Engine

A canonical type system replaces the old per-pair O(n²) mappings. Every engine maps to/from a shared intermediate layer — adding a new database requires only one mapping file.

💻 Platform Support

Datavor now runs on macOS, Windows 10/11, and Linux (Ubuntu 22.04+). The Windows and Linux releases were part of the v2.0 milestone.

Tool Count

Upgrading from v1.5

A history of every Datavor release — what changed, what was fixed, and what's coming next.

🗓️ Scheduler — Automate Your Syncs

You no longer need to be in Claude Desktop for your syncs to run. Create a job once, set a schedule in plain English, and Datavor handles the rest.

Jobs are stored in ~/.datavor/jobs/ — list them, pause them, resume them, or trigger a manual run at any time. Run the scheduler daemon independently for always-on operation.

🔄 Transform Pipeline — Shape Data During Sync

Sync data doesn't always land in the right shape. Column names differ between systems, status codes need expanding, types need converting. Now it all happens inline — no extra passes, no temp tables.

Preview before you commit: transform_preview shows a before/after on real sample data so you can verify before writing anything to the target.

📊 Sync Dashboard — Know What's Running

Every sync is now recorded automatically. Ask Claude for a dashboard at any time — success rates, rows moved, failure diagnostics, and per-table history. Reads from a local ledger file — instant, no database connection needed.

🧠 Type Engine — Smarter Cross-Database Conversion

Replaced best-effort regex type mapping with a proper universal type layer. Every MySQL type maps to a canonical intermediate type, which maps to the correct PostgreSQL equivalent — and vice versa.

🐛 Bug Fixes

Tool Count

Upgrading from v1.0

Initial Release

Change Data Capture (CDC) for real-time sync. dbt integration for transform pipelines.

Local web dashboard for browsing ledger history, managing scheduled jobs, and building transform pipelines visually.