Databases
- PostgreSQL
- MySQL
- MongoDB
11 DB blocks: get, count, exists, insert, upsert, update, delete, transactions, native queries
integrations Connect once. Use from any block.
An integration is a connection your project defines once — credentials, TLS, timeouts — and every block picks it from a list. Secrets live in App Config, never on the canvas.
Read the docsKey-value stores
- Redis
- Memcached
- Valkey
- DragonflyDB
KV Operations (get · set · delete with TTL), KV Raw for anything else
Message queues
- Redis Streams
- RabbitMQ
- Kafka · EE
- NATS JetStream · EE
- Amazon SQS · EE
Triggers that start workflows, and the Send Message block
Observability
- Loki
- OpenTelemetry (OTLP)
Cloud Logs block, and the project's logs, traces and metrics
AI models
- OpenAI
- Anthropic
- Google Gemini
- Mistral
- OpenAI-compatible
Power the AI harness that builds on your canvas
Everything else
- Any HTTP API
- npm packages
HTTP Request block, and imports in JS Runner scripts
EE = Enterprise connectors. Everything else is in the Community edition.
one interface
Switch databases without rewriting routes.
Database blocks speak to an adapter, not to a driver. Point the integration at a different database and the same blocks keep working (joins are SQL-only).
safe by default
Credentials stay out of your flows.
-
App Config
store connection strings and keys as project config; blocks pick them by name
-
Encrypted App Config
config values can be stored encrypted; integrations can read their credentials from config keys
-
Test connection
check an integration from the editor before you use it
-
Overrides in tests
swap an integration for a test one, per suite