# Integrations

# Integrations

Integrations connect Curiosity Workspace to external systems for ingestion, synchronization, and workflow automation.

Common categories:

• Data source integrations: import content from storage, databases, and SaaS systems
• API integrations: pull or push data via REST/GraphQL APIs
• Event-driven integrations: react to changes (webhooks, queues)
• Downstream integrations: send signals/actions to other systems (tickets, notifications)

# Choosing the right integration approach

• Use a connector when you need full control over mapping and graph relationships.
• Use a configured integration when the source system is standard and configuration is sufficient.
• Use scheduled tasks when data changes periodically and near-real-time sync is not required.

See Data Integration → Connectors.

# Integration design principles

• Idempotent writes: repeated runs should not create duplicates.
• Stable identity: consistent keys for nodes and edges.
• Incremental updates: sync deltas where possible.
• Observability: log counts, failures, and latency.
• Security: store credentials in a secret manager; minimize token scopes.

# Common patterns

• Ingest → index → enrich

  • ingest structured records
  • index key fields for search
  • apply NLP/embeddings where beneficial

• Retrieve → decide → act

  • retrieve candidates (graph + search)
  • decide using business rules and/or LLM assistance
  • act through an endpoint that writes back or calls external APIs

# Next steps

• Automate periodic sync and maintenance: Scheduled Tasks
• Add custom API logic: Custom Endpoints