Skip to main content

​
Data Storage & Persistence

The system supports multiple data stores to handle both lightweight development workflows and production-scale deployments.
SQLite is used for local development and simple, single-node workflows where minimal setup and portability are required.
PostgreSQL is recommended for production environments, supporting multi-tenant architectures, higher concurrency, and robust data integrity.
The data store persists all critical operational information, including:
Workflow state to enable recovery, retries, and resumable execution
Agent decisions for traceability and debugging
Human approvals to support human-in-the-loop workflows
Tool outputs generated during task execution
Audit logs for compliance, monitoring, and observability
Context and memory snapshots to maintain continuity across agent runs
This persistence layer ensures reliability, transparency, and full lifecycle visibility across all agent workflows.