Connectors Overview
Connectors link GradientHarbor to your databases. Once connected, the platform introspects your schema, builds a searchable catalog, and enables the AI agent and dashboards to query your data directly.
Supported Connectors
┌─────────────────────────────────────────────────────────────┐
│ Data Warehouse / OLAP │
├──────────────┬──────────────────────────────────────────────┤
│ PostgreSQL │ ✅ Fully supported │
│ BigQuery │ ✅ Fully supported │
│ Snowflake │ 🔜 Coming soon │
│ Redshift │ 🔜 Coming soon │
│ Databricks │ 🔜 Coming soon │
│ ClickHouse │ 📋 Planned │
├──────────────┴──────────────────────────────────────────────┤
│ Document / NoSQL │
├──────────────┬──────────────────────────────────────────────┤
│ MongoDB │ ✅ Supported (via DuckDB SQL interface) │
│ Atlas SQL │ ✅ Supported (via ODBC MongoSQL interface) │
├──────────────┴──────────────────────────────────────────────┤
│ Transactional / OLTP │
├──────────────┬──────────────────────────────────────────────┤
│ MySQL │ 🔜 Coming soon │
├──────────────┴──────────────────────────────────────────────┤
│ File-Based │
├──────────────┬──────────────────────────────────────────────┤
│ DuckDB (CSV) │ ✅ Built-in (upload CSVs via chat) │
└──────────────┴──────────────────────────────────────────────┘| Connector | Status | Use Case | Docs |
|---|---|---|---|
| PostgreSQL | ✅ Available | Primary relational databases, analytics on indexed tables | Setup guide |
| BigQuery | ✅ Available | Google Cloud data warehouse, serverless analytics | Setup guide |
| MongoDB | ✅ Available | Document databases via SQL interface | Setup guide |
| DuckDB (CSV) | ✅ Built-in | Upload and query CSV files directly | Setup guide |
| Snowflake | 🔜 Coming soon | Cloud data warehouse with auto-scaling | Details |
| Redshift | 🔜 Coming soon | AWS columnar data warehouse | Details |
| Databricks | 🔜 Coming soon | Lakehouse analytics with Delta Lake | Details |
| MySQL | 🔜 Coming soon | Common transactional database | Details |
| ClickHouse | 📋 Planned | Open-source columnar OLAP engine | Details |
TIP
Don't see your database? Contact us — we're actively expanding connector support.
Connector Lifecycle
Every connector goes through the same lifecycle:
┌──────────────┐ ┌──────────────────┐ ┌────────────────┐ ┌─────────────────┐
│ Create │ │ Initial │ │ Schema │ │ Full Catalog │
│ Connection │───▶│ Introspection │───▶│ Selection │───▶│ Sync │
│ │ │ │ │ │ │ │
│ Enter creds │ │ Discover DBs & │ │ Enable schemas │ │ Columns, FKs, │
│ + validate │ │ schemas │ │ you need │ │ samples, AI desc│
└──────────────┘ └──────────────────┘ └────────────────┘ └─────────────────┘1. Create Connection
Navigate to Administration > DB Connectors and click Add Connection. Enter the connection credentials. GradientHarbor validates credentials before saving — all credentials are encrypted at rest.
2. Initial Introspection
An async background job discovers available databases and schemas. This runs automatically and typically takes a few seconds.
3. Schema Selection
Select which schemas to catalog. Only enabled schemas are introspected during full sync. You can change this at any time.
4. Full Catalog Sync
For each enabled schema, GradientHarbor captures:
- Columns — Name, data type, nullability
- Indexes — Index names and columns
- Foreign keys — Relationships between tables
- Sample data — Preview rows from each table
- Row counts — Approximate table sizes
- AI descriptions — Auto-generated descriptions of tables and columns
Security
┌────────────────────────────────────────────────────┐
│ Security Layers │
├────────────────────────────────────────────────────┤
│ 🔐 Credentials encrypted at rest (Fernet) │
│ 📖 Read-only query execution │
│ 🏢 Row-level security (org isolation) │
│ 🔒 SSL/TLS support for all connections │
└────────────────────────────────────────────────────┘- Queries execute with read-only permissions
- Each organization's connectors are isolated via row-level security
- SSL certificates supported for encrypted connections